{"id":"4e4e0a61-db0b-4e06-a004-187e307eebc1","entityType":"agent","slug":"clawhub-sciminer-pharma-intelligence","name":"Pharma Intelligence","canonicalUrl":"https://www.xpersona.co/agent/clawhub-sciminer-pharma-intelligence","canonicalPath":"/agent/clawhub-sciminer-pharma-intelligence","generatedAt":"2026-10-11T03:54:16.182Z","source":"CLAWHUB","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-10T23:56:50.891Z","emptyReason":null},"description":"In-depth, multi-region pharmaceutical intelligence search and synthesis, plus drug repurposing, target discovery, clinical evidence review, and bioactivity analysis. Use this skill whenever the user asks about drug approvals, clinical trials, regulatory submissions, pipeline assets, patent landscape","descriptionLabel":"Source description","evidenceSummary":"Capability contract not published. No trust telemetry is available yet. 1.2K downloads reported by the source. Last updated 10/10/2026.","installCommand":"clawhub skill install s170qp1q4twz35wa85ppa8894h83w461:pharma-intelligence","sourceUrl":"https://clawhub.ai/sciminer/pharma-intelligence","homepage":"https://clawhub.ai/sciminer/skills/pharma-intelligence","primaryLinks":[{"label":"View on ClawHub","url":"https://clawhub.ai/sciminer/pharma-intelligence","kind":"source"},{"label":"Homepage","url":"https://clawhub.ai/sciminer/skills/pharma-intelligence","kind":"homepage"}],"safetyScore":84,"overallRank":62,"popularityScore":62,"trustScore":null,"claimedByName":null,"isOwner":false,"seoDescription":"Pharma Intelligence technical dossier on Xpersona with agent coverage, OPENCLEW support, and live trust metadata."},"coverage":{"evidence":{"source":"public-profile","verified":false,"confidence":"medium","updatedAt":"2026-10-10T23:56:50.891Z","emptyReason":null},"protocols":[{"protocol":"OPENCLEW","label":"OpenClaw","status":"self-declared","notes":"Declared in the public agent profile."}],"capabilities":[],"verifiedCount":0,"selfDeclaredCount":1,"capabilityMatrix":{"rows":[{"key":"OPENCLEW","type":"protocol","support":"unknown","confidenceSource":"profile","notes":"Listed on profile"}],"flattenedTokens":"protocol:OPENCLEW|unknown|profile"}},"adoption":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-10T23:56:50.891Z","emptyReason":null},"stars":null,"forks":null,"downloads":1225,"packageName":null,"latestVersion":"1.0.3","tractionLabel":"1.2K downloads"},"release":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-10T23:56:50.826Z","emptyReason":null},"lastUpdatedAt":"2026-10-10T23:56:50.891Z","lastCrawledAt":"2026-10-10T23:56:50.826Z","lastIndexedAt":null,"nextCrawlAt":"2026-10-11T23:56:50.826Z","lastVerifiedAt":null,"highlights":[{"version":"1.0.3","createdAt":"2026-07-06T15:14:37.496Z","changelog":"pharma-intelligence 1.0.3 - Removed the skill-card.md file. - SKILL.md updated to clarify tool usage: always try web_fetch first for all sources; if blocked or empty, retry with browser_navigate before concluding no data. - Expanded and revised list of biomedical research databases covered (now 20+). - Added a new \"Tool Access Notes\" section detailing the web_fetch/browser_navigate fallback rule for better source reliability. - Per-intent workflows now explicitly defer to the new tool fallback rule, increasing guidance consistency.","fileCount":50,"zipByteSize":112907},{"version":"1.0.2","createdAt":"2026-06-09T07:37:59.855Z","changelog":"**Major update: pharma-intelligence skill now re-architected for modular sub-skill execution.** - Introduced modular \"sub-skills\" for each key biomedical/pharma data source, each with dedicated scripts and invocation instructions. - Removed reliance on a single MCP server endpoint; now delegates research tasks to specific local sub-skills (e.g., clinicaltrials-skill, chembl-skill, ncbi-entrez-skill). - Updated workflows and invocation instructions throughout SKILL.md to reference these sub-skills and their new interfaces. - Added references/sub-skills.md to map research tasks to sub-skills and clarify execution patterns. - Clarified ID resolution steps and data source usage for each search intent. - Improved documentation, removed obsolete MCP instructions, and adjusted per-intent research patterns to leverage new sub-skill structure.","fileCount":50,"zipByteSize":111357},{"version":"1.0.1","createdAt":"2026-05-11T08:55:11.745Z","changelog":"pharma-intelligence 1.0.1 - Search workflow and per-intent sequences have been moved to a dedicated reference file ([references/pharma-intelligence-workflow.md]). - Streamlined SKILL.md: now refers to newly added reference documentation for workflows and regional source maps. - Updated MCP server instructions: users must call the endpoint directly via HTTP POST using web_fetch or run_in_terminal; no built-in MCP tool. - Approval/trial workflows now emphasize explicit, step-by-step instructions and highlight which source to use when MCP coverage is lacking. - Centralized region-by-region source list and workflow examples into referenced documentation for easier updates and clarity.","fileCount":8,"zipByteSize":23034},{"version":"1.0.0","createdAt":"2026-05-11T05:52:17.623Z","changelog":"pharma-intelligence 1.0.0 - Initial release of pharma-intelligence skill for in-depth, multi-region pharmaceutical intelligence search and synthesis. - Provides systematic, tiered-source searching for drug approvals, clinical trials, regulatory submissions, pipelines, patents, targets, and bioactivity across major global regions (China, US, Europe, Japan, South Korea, Australia). - Integrates regulatory, clinical, academic, and commercial databases using a unified MCP server endpoint. - Details a 3-tier source hierarchy and region-specific database/source maps. - Includes recommended tool usage, example queries, and region-by-region data acquisition strategies. - Ensures source-grounded responses for any pharma, drug, target, or biomedical research questions.","fileCount":7,"zipByteSize":23988}]},"execution":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No published capability contract is available yet."},"installCommand":"clawhub skill install s170qp1q4twz35wa85ppa8894h83w461:pharma-intelligence","setupComplexity":"low","setupSteps":["Install using `clawhub skill install s170qp1q4twz35wa85ppa8894h83w461:pharma-intelligence` in an isolated environment before connecting it to live workloads.","No published capability contract is available yet, so validate auth and request/response behavior manually.","Review the upstream CLAWHUB listing at https://clawhub.ai/sciminer/pharma-intelligence before using production credentials."],"contract":{"contractStatus":"missing","authModes":[],"requires":[],"forbidden":[],"supportsMcp":false,"supportsA2a":false,"supportsStreaming":false,"inputSchemaRef":null,"outputSchemaRef":null,"dataRegion":null,"contractUpdatedAt":null,"sourceUpdatedAt":null,"freshnessSeconds":null},"invocationGuide":{"preferredApi":{"snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/trust"},"curlExamples":["curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/snapshot\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/contract\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/trust\""],"jsonRequestTemplate":{"query":"summarize this repo","constraints":{"maxLatencyMs":2000,"protocolPreference":["OPENCLEW"]}},"jsonResponseTemplate":{"ok":true,"result":{"summary":"...","confidence":0.9},"meta":{"source":"CLAWHUB","generatedAt":"2026-10-11T03:54:16.176Z"}},"retryPolicy":{"maxAttempts":3,"backoffMs":[500,1500,3500],"retryableConditions":["HTTP_429","HTTP_503","NETWORK_TIMEOUT"]}},"endpoints":{"dossierUrl":"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/dossier","snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-sciminer-pharma-intelligence/trust"}},"reliability":{"evidence":{"source":"runtime-metrics","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No trust, reliability, or runtime telemetry is available."},"trust":{"status":"unavailable","handshakeStatus":"UNKNOWN","verificationFreshnessHours":null,"reputationScore":null,"p95LatencyMs":null,"successRate30d":null,"fallbackRate":null,"attempts30d":null,"trustUpdatedAt":null,"trustConfidence":"unknown","sourceUpdatedAt":null,"freshnessSeconds":null},"decisionGuardrails":{"doNotUseIf":["Contract metadata is missing or unavailable for deterministic execution."],"safeUseWhen":[],"riskFlags":["missing_or_unavailable_contract","trust_data_unavailable","schema_references_missing"],"operationalConfidence":"low"},"executionMetrics":{"observedLatencyMsP50":null,"observedLatencyMsP95":null,"estimatedCostUsd":null,"uptime30d":null,"rateLimitRpm":null,"rateLimitBurst":null,"lastVerifiedAt":null,"verificationSource":null},"runtimeMetrics":{"successRate":null,"avgLatencyMs":null,"avgCostUsd":null,"hallucinationRate":null,"retryRate":null,"disputeRate":null,"p50Latency":null,"p95Latency":null,"lastUpdated":null}},"benchmarks":{"evidence":{"source":"no-benchmark-data","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No benchmark suites or observed failure patterns are available."},"suites":[],"failurePatterns":[]},"artifacts":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-10T23:56:50.891Z","emptyReason":null},"readme":"Skill: Pharma Intelligence\n\nOwner: sciminer\n\nSummary: In-depth, multi-region pharmaceutical intelligence search and synthesis, plus drug repurposing, target discovery, clinical evidence review, and bioactivity analysis. Use this skill whenever the user asks about drug approvals, clinical trials, regulatory submissions, pipeline assets, patent landscape\n\nTags: latest:1.0.3\n\nVersion history:\n\nv1.0.3 | 2026-07-06T15:14:37.496Z | user\n\npharma-intelligence 1.0.3\n\n- Removed the skill-card.md file.\n- SKILL.md updated to clarify tool usage: always try web_fetch first for all sources; if blocked or empty, retry with browser_navigate before concluding no data.\n- Expanded and revised list of biomedical research databases covered (now 20+).\n- Added a new \"Tool Access Notes\" section detailing the web_fetch/browser_navigate fallback rule for better source reliability.\n- Per-intent workflows now explicitly defer to the new tool fallback rule, increasing guidance consistency.\n\nv1.0.2 | 2026-06-09T07:37:59.855Z | user\n\n**Major update: pharma-intelligence skill now re-architected for modular sub-skill execution.**\n\n- Introduced modular \"sub-skills\" for each key biomedical/pharma data source, each with dedicated scripts and invocation instructions.\n- Removed reliance on a single MCP server endpoint; now delegates research tasks to specific local sub-skills (e.g., clinicaltrials-skill, chembl-skill, ncbi-entrez-skill).\n- Updated workflows and invocation instructions throughout SKILL.md to reference these sub-skills and their new interfaces.\n- Added references/sub-skills.md to map research tasks to sub-skills and clarify execution patterns.\n- Clarified ID resolution steps and data source usage for each search intent.\n- Improved documentation, removed obsolete MCP instructions, and adjusted per-intent research patterns to leverage new sub-skill structure.\n\nv1.0.1 | 2026-05-11T08:55:11.745Z | user\n\npharma-intelligence 1.0.1\n\n- Search workflow and per-intent sequences have been moved to a dedicated reference file ([references/pharma-intelligence-workflow.md]).\n- Streamlined SKILL.md: now refers to newly added reference documentation for workflows and regional source maps.\n- Updated MCP server instructions: users must call the endpoint directly via HTTP POST using web_fetch or run_in_terminal; no built-in MCP tool.\n- Approval/trial workflows now emphasize explicit, step-by-step instructions and highlight which source to use when MCP coverage is lacking.\n- Centralized region-by-region source list and workflow examples into referenced documentation for easier updates and clarity.\n\nv1.0.0 | 2026-05-11T05:52:17.623Z | user\n\npharma-intelligence 1.0.0\n\n- Initial release of pharma-intelligence skill for in-depth, multi-region pharmaceutical intelligence search and synthesis.\n- Provides systematic, tiered-source searching for drug approvals, clinical trials, regulatory submissions, pipelines, patents, targets, and bioactivity across major global regions (China, US, Europe, Japan, South Korea, Australia).\n- Integrates regulatory, clinical, academic, and commercial databases using a unified MCP server endpoint.\n- Details a 3-tier source hierarchy and region-specific database/source maps.\n- Includes recommended tool usage, example queries, and region-by-region data acquisition strategies.\n- Ensures source-grounded responses for any pharma, drug, target, or biomedical research questions.\n\nArchive index:\n\nArchive v1.0.3: 50 files, 112907 bytes\n\nFiles: references/drug-naming.md (4200b), references/pharma-intelligence-workflow.md (8919b), references/regulatory-timelines.md (5274b), references/sources-by-region.md (8667b), references/sub-skills.md (8873b), skill-card.md (2866b), SKILL.md (16697b), skills/biorxiv-skill/scripts/rest_request.py (11019b), skills/biorxiv-skill/SKILL.md (2555b), skills/chebi-skill/scripts/rest_request.py (11019b), skills/chebi-skill/SKILL.md (2216b), skills/chembl-skill/scripts/rest_request.py (11019b), skills/chembl-skill/SKILL.md (2505b), skills/clinicaltrials-skill/scripts/clinicaltrials_client.py (9026b), skills/clinicaltrials-skill/SKILL.md (2426b), skills/clinvar-variation-skill/scripts/clinvar_variation.py (8028b), skills/clinvar-variation-skill/SKILL.md (2050b), skills/efo-ontology-skill/scripts/rest_request.py (11019b), skills/efo-ontology-skill/SKILL.md (2450b), skills/ensembl-skill/scripts/rest_request.py (11019b), skills/ensembl-skill/SKILL.md (2407b), skills/gnomad-graphql-skill/scripts/gnomad_graphql.py (5770b), skills/gnomad-graphql-skill/SKILL.md (2147b), skills/gwas-catalog-skill/scripts/rest_request.py (11019b), skills/gwas-catalog-skill/SKILL.md (2605b), skills/ncbi-clinicaltables-skill/scripts/ncbi_gene_clinicaltables.py (5705b), skills/ncbi-clinicaltables-skill/SKILL.md (2410b), skills/ncbi-entrez-skill/references/geo.md (952b), skills/ncbi-entrez-skill/scripts/ncbi_entrez.py (11130b), skills/ncbi-entrez-skill/SKILL.md (2773b), skills/ncbi-pmc-skill/scripts/ncbi_pmc.py (9011b), skills/ncbi-pmc-skill/SKILL.md (1845b), skills/opentargets-skill/scripts/opentargets_disease_heatmap.py (12448b), skills/opentargets-skill/scripts/opentargets_graphql.py (5800b), skills/opentargets-skill/SKILL.md (3297b), skills/pubchem-pug-skill/scripts/rest_request.py (11019b), skills/pubchem-pug-skill/SKILL.md (2405b), skills/quickgo-skill/scripts/rest_request.py (11019b), skills/quickgo-skill/SKILL.md (2497b), skills/reactome-skill/scripts/rest_request.py (11019b), skills/reactome-skill/SKILL.md (2435b), skills/rhea-skill/scripts/rest_request.py (11019b), skills/rhea-skill/SKILL.md (2027b), skills/rnacentral-skill/scripts/rest_request.py (11019b), skills/rnacentral-skill/SKILL.md (2248b), skills/string-skill/scripts/rest_request.py (11019b), skills/string-skill/SKILL.md (2596b), skills/uniprot-skill/scripts/rest_request.py (11019b), skills/uniprot-skill/SKILL.md (2774b), _meta.json (138b)\n\nFile v1.0.3:SKILL.md\n\n---\r\nname: pharma-intelligence\r\ndescription:\r\n  In-depth, multi-region pharmaceutical intelligence search and synthesis,\r\n  plus drug repurposing, target discovery, clinical evidence review, and\r\n  bioactivity analysis. Use this skill whenever the user asks about drug\r\n  approvals, clinical trials, regulatory submissions, pipeline assets, patent\r\n  landscapes, competitive intelligence, scientific evidence, disease targets,\r\n  genetic associations, or compound bioactivity for any drug, target,\r\n  indication, or company — especially when coverage of China, US, Europe,\r\n  Japan, South Korea, or Australia is needed. Trigger even for casual queries\r\n  like \"what's the approval status of X in China\", \"find trials for Y in\r\n  Japan\", \"compare pipeline coverage across regions\", \"find drugs for disease\r\n  Z\", or \"what targets are associated with condition W\". Always consult this\r\n  skill before answering any pharma or biomedical research question that\r\n  requires source-grounded data.\r\n---\r\n\r\n# Global Pharma Intelligence & Biomedical Research Skill\r\n\r\nSystematic, source-prioritized search and synthesis across regulatory, clinical,\r\nacademic, and commercial databases — covering all major pharmaceutical markets\r\nand 20+ biomedical research databases.\r\n\r\n## Sub-Skills — How to Invoke\r\n\r\nThis skill delegates all database work to the sub-skills bundled locally under `skills/`.\r\nRead the relevant sub-skill's `SKILL.md` before invoking it, then run its bundled script.\r\n\r\nSee [references/sub-skills.md](./references/sub-skills.md) for the full mapping of research\r\ntasks to sub-skills and execution patterns.\r\n\r\n---\r\n\r\n## Core Principle: Tiered Source Priority\r\n\r\nEvery region follows a 3-tier hierarchy. Higher tiers override lower-tier claims; always cite the tier.\r\n\r\n| Tier | Type | Description |\r\n|------|------|-------------|\r\n| **Tier 1** | Regulatory | Official agency submissions, approvals, labels |\r\n| **Tier 2** | Trial registries | Prospective/registered clinical evidence |\r\n| **Tier 3** | Academic / IP | Published papers, conferences, patents |\r\n\r\nFor the per-region source map (CN / US / EU / JP / KR / AU + global) with URLs and access notes, see [references/sources-by-region.md](./references/sources-by-region.md).\r\n\r\n---\r\n\r\n## Tool Access Notes\r\n\r\n`web_fetch` is the default tool for any URL in this skill that isn't covered by a bundled sub-skill. Some sites are JavaScript-rendered or block plain HTTP fetches — Google Patents is the most common offender, and CTIS, jRCT, and ANZCTR occasionally behave the same way — but this can happen on **any** site, not just those.\r\n\r\n**Rule:** try `web_fetch` first. If it returns empty, blocked, or placeholder content, retry the exact same URL with `browser_navigate` before concluding that a source has no data. Every other section in this skill that mentions `web_fetch` defers to this rule rather than restating it.\r\n\r\n---\r\n\r\n## Search Workflow\r\n\r\n### Step 1 — Classify the Query (pick ONE intent)\r\n\r\n| # | Intent | Trigger phrases |\r\n|---|--------|-----------------|\r\n| A | Trial landscape | \"trials of X\", \"clinical studies of\", \"who is testing\", \"phase 2/3 of\" |\r\n| B | Approval / regulatory status | \"is X approved\", \"approval status\", \"FDA/EMA/NMPA cleared\" |\r\n| C | Safety / adverse events | \"side effects of\", \"is X safe\", \"adverse events\", \"black box\" |\r\n| D | Pipeline / competitive intel | \"pipeline\", \"competitive landscape\", \"who else is developing\" |\r\n| E | Patent / IP / exclusivity | \"when does patent expire\", \"patent landscape\", \"exclusivity\" |\r\n| F | Target / mechanism / drug discovery | \"drugs targeting X\", \"mechanism of\", \"bioactivity\", \"IC50\" |\r\n| G | Repurposing / target discovery | \"repurpose for\", \"targets associated with disease\", \"genetic basis\" |\r\n| H | Literature / evidence review | \"recent papers on\", \"what's known about\", \"systematic review\" |\r\n\r\nAlso capture: **regions in scope** (US / EU / JP / CN / KR / AU / global) and **time horizon**.\r\n\r\n### Step 2 — Execute the Per-Intent Sequence\r\n\r\nRun the workflow for the chosen intent (see [Per-Intent Workflows](#per-intent-workflows)) in order. For sources without MCP coverage (CN NMPA/CDE, EMA EPAR, PMDA, jRCT, CTIS, CRIS, ANZCTR, Orange Book), use `web_fetch` only at the steps that name them, following the fallback rule in [Tool Access Notes](#tool-access-notes).\r\n\r\nResolve identifiers as needed:\r\n- Free-text disease → MONDO/EFO ID via `opentargets-skill` or `efo-ontology-skill`\r\n- Free-text gene → HGNC symbol via `ncbi-clinicaltables-skill` or `ensembl-skill`\r\n- Cross-database ID conversion → `ensembl-skill`, `uniprot-skill`, or `efo-ontology-skill`\r\n\r\n### Step 3 — Resolve Conflicts\r\n\r\n1. Higher-tier source wins (Tier 1 > Tier 2 > Tier 3).\r\n2. More recent data wins within the same tier.\r\n3. Flag unresolved conflicts; do not silently pick one.\r\n\r\n### Step 4 — Synthesize and Present\r\n\r\nStructure output to match the intent of the question:\r\n- Trial landscape → table of trials (NCT/registry ID, phase, status, sponsor, N, primary endpoint).\r\n- Approval status → region × status × date × indications table.\r\n- Safety → top FAERS reactions plus black-box / warnings.\r\n- Pipeline → drug × company × phase × mechanism table.\r\n- Patent → patent number, jurisdiction, expiry.\r\n\r\nAlways cite source, tier, and access date.\r\n\r\n---\r\n\r\n## Per-Intent Workflows\r\n\r\n### A. Trial Landscape\r\n\r\n*\"What clinical studies / trials exist for [drug | target | indication]?\"*\r\n\r\nDefault scope = ALL regions. Only narrow if the user names a single region.\r\n\r\n   `clinicaltrials-skill` covers only ClinicalTrials.gov, which is primarily US-registered trials. Run each regional source in parallel.\r\n\r\n1. **United States** — `clinicaltrials-skill` (`action=studies`).\r\n   - Use `query.intr` for a drug, `query.cond` for a disease, both for combined.\r\n   - For a target/class (e.g., \"pan-RAS\", \"PD-L1 inhibitor\"): pass the class term as `query.intr` plus a relevant `query.cond`.\r\n   - Then re-run with an NCT ID in `query.id` for eligibility, endpoints, sponsor, and locations.\r\n2. **China** — `web_fetch`:\r\n   - `http://www.chinadrugtrials.org.cn` (mandatory CN IND registry)\r\n   - `https://www.chictr.org.cn` (ChiCTR, WHO primary)\r\n3. **Europe** — `web_fetch`:\r\n   - `https://euclinicaltrials.eu` (CTIS — current EU register)\r\n   - `https://eudract.ema.europa.eu` (EudraCT — legacy historical trials)\r\n   - `https://www.isrctn.com` (ISRCTN, UK/global)\r\n4. **Japan** — `web_fetch`:\r\n   - `https://jrct.niph.go.jp` (jRCT — mandatory JP registry)\r\n   - `https://www.umin.ac.jp/ctr/` (UMIN-CTR — legacy)\r\n5. **South Korea** — `web_fetch` `https://cris.nih.go.kr`.\r\n6. **Australia / New Zealand** — `web_fetch` `https://www.anzctr.org.au`.\r\n7. **WHO ICTRP catch-all** — `web_fetch` `https://trialsearch.who.int` for any WHO primary registry (covers India CTRI, Iran IRCT, Brazil ReBEC, etc.).\r\n8. **Published results** — `ncbi-entrez-skill` (`db=pubmed`) with NCT ID or drug name to surface completed-trial papers.\r\n9. **US company-disclosed pipeline** (optional) — `web_fetch` SEC EDGAR full-text search at `https://efts.sec.gov/LATEST/search-index` for US-listed sponsors.\r\n\r\nFor every regional `web_fetch`: query both INN and brand name; for CN also use the Chinese transliteration (see [references/drug-naming.md](./references/drug-naming.md)). Aggregate results in one table with a \"Registry\" column.\r\n\r\n### B. Approval / Regulatory Status\r\n\r\n*\"Is [drug] approved in [region]?\"*\r\n\r\n1. **US** — `web_fetch` `https://api.fda.gov/drug/label.json` (openFDA) and `https://dailymed.nlm.nih.gov/dailymed/services/v2/spls.json` (label date anchors approval); `web_fetch` `https://api.fda.gov/drug/ndc.json` for orphan status.\r\n2. **Non-US** — `web_fetch` the regional Tier 1 source (NMPA, EMA EPAR, PMDA, MFDS, TGA). For CN, also search Chinese characters.\r\n3. `chembl-skill` (`drug_indication.json?molecule_chembl_id=...`) — cross-check approved indications and max phase.\r\n4. Say \"not approved\" only when Tier 1 affirms denial/withdrawal. Otherwise: \"no record found as of [date]\".\r\n\r\n### C. Safety / Adverse Events\r\n\r\n1. `web_fetch` `https://api.fda.gov/drug/event.json` (FAERS) filtering by drug name and seriousness.\r\n2. `web_fetch` `https://api.fda.gov/drug/label.json` with `sections=warnings` and `sections=contraindications`.\r\n3. `chembl-skill` (`molecule/<id>.json`) — inspect the `black_box_warning` flag.\r\n4. `ncbi-entrez-skill` (`db=pubmed`) with terms `\"adverse effect\" OR \"toxicity\"` for case reports and post-marketing literature.\r\n\r\n### D. Pipeline / Competitive Intelligence\r\n\r\n*\"Who else is developing for [indication / target]? What's the global competitive landscape?\"*\r\n\r\nDefault scope = ALL regions. A competitive landscape without the active-trial picture is incomplete, so run the full multi-region trial sweep from Workflow A and then layer pipeline-specific sources on top.\r\n\r\n1. **Active trials — all regions** — run [Workflow A](#a-trial-landscape) steps 1–7 in full, optionally adding `filter.overallStatus=RECRUITING` (or `ACTIVE_NOT_RECRUITING`) and a phase filter to focus on competitors at a specific stage.\r\n2. **Company disclosures** — `web_fetch` SEC EDGAR full-text search at `https://efts.sec.gov/LATEST/search-index` for pipeline language in 10-K / 10-Q / 8-K (US-listed sponsors only).\r\n3. **Patent activity per company** — `web_fetch` `https://patents.google.com/?assignee=companyname` (see Workflow E, step 1, for the exact query format and non-English name handling; or use WIPO PATENTSCOPE / Espacenet as alternatives).\r\n4. **Published results** — `ncbi-entrez-skill` (`db=pubmed`) with NCT IDs or drug names to surface completed-trial papers.\r\n\r\nAggregate into one table: drug × company × phase × mechanism × registry/region.\r\n\r\n### E. Patent / IP / Exclusivity\r\n\r\nAll listed patent sources are free and require no API key.\r\n\r\n1. **Global patent search** — `web_fetch` one or more of (see [Tool Access Notes](#tool-access-notes) for the `browser_navigate` fallback):\r\n   - `https://patents.google.com` (Google Patents — best full-text search, covers USPTO, EPO, WIPO, JPO, CNIPA, KIPO). To retrieve all patents assigned to a specific company, query `https://patents.google.com/?assignee=companyname`. If the company's name is non-English, first search with the original non-English name, then run a second search with the English translation/transliteration — assignee records are not always normalized across languages, so neither search alone is reliable.\r\n   - `https://patentscope.wipo.int` (WIPO PATENTSCOPE — authoritative for PCT applications and national filings worldwide).\r\n   - `https://worldwide.espacenet.com` (EPO Espacenet — strongest European and family-tree coverage).\r\n2. **US patents (structured)** — `uspto_ppubs_search_patents` via MCP for granted patents and applications.\r\n3. **Patent family / cross-jurisdiction equivalents** — Espacenet's \"INPADOC patent family\" view, or Google Patents' \"Worldwide applications\" section.\r\n4. **Orange Book** (patent + exclusivity expiry for FDA-approved drugs) — `web_fetch` `https://www.accessdata.fda.gov/scripts/cder/ob`.\r\n5. **Orphan exclusivity** — `fda_orphan_search_exclusivity` (7-year US orphan exclusivity).\r\n\r\n### F. Target / Mechanism / Drug Discovery\r\n\r\n1. `chembl-skill` — search `target/search.json?q=<gene>` to resolve target ChEMBL ID, then `mechanism.json?target_chembl_id=...` for all drugs.\r\n2. `chembl-skill` — `mechanism.json?molecule_chembl_id=...` for mechanism of action of each candidate.\r\n3. `chembl-skill` — `activity.json?target_chembl_id=...` for IC50 / Kd / EC50 bioactivity comparisons.\r\n4. `uniprot-skill` — `uniprotkb/search` with `gene:<symbol> AND organism_id:9606` for protein function and druggability context.\r\n5. `reactome-skill` — pathway and disease-pathway context for the target.\r\n\r\n### G. Repurposing / Target Discovery\r\n\r\n1. `opentargets-skill` — search for disease to resolve MONDO / EFO ID.\r\n2. `opentargets-skill` — `associatedTargets` query with disease EFO ID → ranked targets by evidence score.\r\n3. `gwas-catalog-skill` — associations for the disease EFO term to identify genetically supported targets.\r\n4. `web_fetch` OMIM API at `https://api.omim.org/api/entry/search` for Mendelian basis (requires API key).\r\n5. For each top target: `chembl-skill` — `mechanism.json?target_chembl_id=...` for all drugs.\r\n6. `clinicaltrials-skill` with each drug as `query.intr` for prior-art trials.\r\n7. `web_fetch` `https://api.fda.gov/drug/event.json` as a safety filter for non-trivial candidates.\r\n\r\n### H. Literature / Evidence Review\r\n\r\n1. `ncbi-entrez-skill` (`db=pubmed`) — entry point; use MeSH terms for disease, chemical, and gene-aware filtering.\r\n2. `web_fetch` Europe PMC REST (`https://www.ebi.ac.uk/europepmc/webservices/rest/search`) — broader: grants, preprints, non-MEDLINE.\r\n3. `biorxiv-skill` — bioRxiv / medRxiv preprints only.\r\n4. `ncbi-pmc-skill` or `ncbi-entrez-skill` (`efetch`, `db=pmc`) — abstract or full text for top hits.\r\n\r\n---\r\n\r\n## Combination Strategies (cross-intent)\r\n\r\nUse only when a question genuinely spans multiple intents.\r\n\r\n- **Disease → Targets → Drugs → Trials**: `opentargets-skill` (search + associations) → `chembl-skill` (mechanism by target) → `clinicaltrials-skill`\r\n- **Gene → Protein → Pathways → Drugs**: `ncbi-clinicaltables-skill` or `ensembl-skill` → `uniprot-skill` → `reactome-skill` → `chembl-skill` (mechanism by target)\r\n- **Variant → Gene → Disease → Treatments**: `clinvar-variation-skill` or `gnomad-graphql-skill` → `ensembl-skill` → `opentargets-skill` → `chembl-skill` (mechanism by target)\r\n- **Drug → Safety → Label → Trials**: `chembl-skill` (mechanism) → `web_fetch` openFDA adverse events → `web_fetch` openFDA label → `clinicaltrials-skill`\r\n\r\n---\r\n\r\n## API Keys\r\n\r\nMost APIs require no key. Exceptions:\r\n\r\n| Database | Key | Source |\r\n|----------|-----|--------|\r\n| OMIM | Required | https://omim.org/api |\r\n| NCI Clinical Trials | Optional | https://clinicaltrialsapi.cancer.gov |\r\n| OpenFDA | Optional (higher rate limits) | https://open.fda.gov/apis |\r\n\r\nAll bundled sub-skills (ChEMBL, OpenTargets, PubMed via NCBI Entrez, ClinicalTrials.gov, Reactome, UniProt, GWAS Catalog, Ensembl) are public and require no key. Patent landscape work uses Google Patents, WIPO PATENTSCOPE, and Espacenet — no keys required.\r\n\r\n---\r\n\r\n## Output Quality Standards\r\n\r\n- Never fabricate approval dates, trial IDs, or efficacy numbers.\r\n- Attribute every claim to its source and tier.\r\n- Flag gaps explicitly (e.g., \"No registered trials found in jRCT as of [date]\").\r\n- Distinguish \"no data found\" from \"not approved\" — absence of evidence ≠ negative regulatory decision.\r\n- For Chinese sources: note whether the search was conducted in Chinese characters; romanization alone may miss records.\r\n\r\n---\r\n\r\n## Troubleshooting\r\n\r\n**No results?**\r\n- Try alternative terms (INN vs brand name, gene symbol vs protein name).\r\n- Use standardized IDs: MONDO/EFO for diseases, HGNC for genes, ChEMBL IDs for compounds, Ensembl for OpenTargets.\r\n- Resolve IDs first with `efo-ontology-skill`, `ncbi-clinicaltables-skill`, `ensembl-skill`, or `uniprot-skill`.\r\n\r\n**Too many results?**\r\n- Add filters: `max_items`, `filter.phase`, `filter.overallStatus`, `reviewed=true` (UniProt).\r\n- Apply date ranges where supported.\r\n\r\n**API key errors?**\r\n- OMIM requires a key; NCI and OpenFDA accept optional keys for higher rate limits.\r\n\r\n**Source not covered by a sub-skill?**\r\n- Use `web_fetch` directly for CDE/NMPA, EMA/EPAR, PMDA, jRCT, CTIS, CRIS, ANZCTR, Orange Book, openFDA, DailyMed, FAERS, and EDGAR.\r\n\r\n**`web_fetch` returns empty, blocked, or placeholder content?**\r\n- See [Tool Access Notes](#tool-access-notes): retry the same URL with `browser_navigate` before concluding the source has no data.\r\n\r\n---\r\n\r\n## References\r\n\r\n- [references/sub-skills.md](./references/sub-skills.md) — Mapping of pharma-intelligence tasks to bundled sub-skills, with execution patterns.\r\n- [references/drug-naming.md](./references/drug-naming.md) — INN / brand / Chinese / Japanese naming conventions and transliteration.\r\n- [references/regulatory-timelines.md](./references/regulatory-timelines.md) — Review-clock lengths and milestones per agency (FDA, EMA, PMDA, CDE/NMPA, etc.).\r\n- [references/sources-by-region.md](./references/sources-by-region.md) — Direct URLs and access notes for all regional regulatory databases.\r\n- [references/pharma-intelligence-workflow.md](./references/pharma-intelligence-workflow.md) — End-to-end worked example (osimertinib in NSCLC).\n\nFile v1.0.3:skills/biorxiv-skill/SKILL.md\n\n---\r\nname: biorxiv-skill\r\ndescription: Submit compact bioRxiv and medRxiv API requests for details, publication-linkage, and DOI lookups. Use when a user wants concise preprint metadata summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/rest_request.py` for all bioRxiv and medRxiv API calls.\r\n- Use `base_url=https://api.biorxiv.org`.\r\n- The script accepts `max_items`; for `details` and `pubs` pages, start around `max_items=10`.\r\n- Prefer one cursor page at a time instead of increasing page size or pasting long collections into chat.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not part of the true request.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return the raw script JSON only if the user explicitly asks for machine-readable output.\r\n- Prefer these paths: `details/<server>/<start>/<end>/<cursor>/json`, `details/<server>/<doi>/na/json`, `pubs/<server>/<start>/<end>/<cursor>`, and `pubs/<server>/<doi>/na/json`.\r\n- If the user needs full page contents, set `save_raw=true` and report the saved file path rather than pasting large collections into chat.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required fields: `base_url`, `path`\r\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common biorxiv patterns:\r\n  - `{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/biorxiv/2025-03-21/2025-03-28/0/json\",\"record_path\":\"collection\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/medrxiv/10.1101/2020.09.09.20191205/na/json\",\"record_path\":\"collection\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"pubs/medrxiv/2020-03-01/2020-03-30/0\",\"record_path\":\"collection\",\"max_items\":10}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/biorxiv/2025-03-21/2025-03-28/0/json\",\"record_path\":\"collection\",\"max_items\":10}' | python scripts/rest_request.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.3:skills/chebi-skill/SKILL.md\n\n---\r\nname: chebi-skill\r\ndescription: Submit compact ChEBI 2.0 API requests for chemical search, compound lookup, ontology traversal, and structure metadata. Use when a user wants concise ChEBI summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/rest_request.py` for all ChEBI calls.\r\n- Use `base_url=https://www.ebi.ac.uk`.\r\n- Prefer the documented public routes under `chebi/backend/api/public/`.\r\n- Start with `es_search/` for free-text lookup and use `compound/<CHEBI:id>/` for targeted records.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return raw JSON only if the user explicitly asks for machine-readable output.\r\n- Prefer these paths: `chebi/backend/api/public/es_search/`, `chebi/backend/api/public/compound/<CHEBI:id>/`, and ontology child or parent routes.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required fields: `base_url`, `path`\r\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common ChEBI patterns:\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/es_search/\",\"params\":{\"query\":\"caffeine\",\"size\":10},\"record_path\":\"results\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/compound/CHEBI:27732/\"}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/ontology/children/CHEBI:27732/\"}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/es_search/\",\"params\":{\"query\":\"caffeine\",\"size\":10},\"record_path\":\"results\",\"max_items\":10}' | python scripts/rest_request.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.3:skills/chembl-skill/SKILL.md\n\n---\r\nname: chembl-skill\r\ndescription: Submit compact ChEMBL API requests for activity, molecule, target, mechanism, and text-search endpoints. Use when a user wants concise ChEMBL summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/rest_request.py` for all ChEMBL API calls.\r\n- Use `base_url=https://www.ebi.ac.uk/chembl/api/data`.\r\n- The script accepts `max_items`; for activity, mechanism, and text-search collections, start with API `limit=10` and `max_items=10`.\r\n- Single molecule or target lookups usually do not need `max_items`.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return the script JSON verbatim only if the user explicitly asks for machine-readable output.\r\n- Prefer these paths: `activity.json`, `molecule/<id>.json`, `target/<id>.json`, `mechanism.json`, and `molecule/search.json`.\r\n- Use `record_path` to target list fields like `activities`, `mechanisms`, or `molecules`.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required fields: `base_url`, `path`\r\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common ChEMBL patterns:\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"activity.json\",\"params\":{\"molecule_chembl_id\":\"CHEMBL25\",\"limit\":10},\"record_path\":\"activities\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"molecule/CHEMBL25.json\"}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"molecule/search.json\",\"params\":{\"q\":\"imatinib\",\"limit\":10},\"record_path\":\"molecules\",\"max_items\":10}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"activity.json\",\"params\":{\"molecule_chembl_id\":\"CHEMBL25\",\"limit\":10},\"record_path\":\"activities\",\"max_items\":10}' | python scripts/rest_request.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.3:skills/clinicaltrials-skill/SKILL.md\n\n---\r\nname: clinicaltrials-skill\r\ndescription: Submit compact ClinicalTrials.gov API v2 requests for study search, metadata, enums, search areas, and field statistics. Use when a user wants concise ClinicalTrials.gov summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/clinicaltrials_client.py` for all ClinicalTrials.gov v2 calls.\r\n- Study searches are better with `max_items=10` and `max_pages=1`; only increase pages when the user explicitly wants more than the first page.\r\n- Use targeted `params` instead of broad unfiltered study dumps.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Prefer `action=studies` for search and `action=metadata|search_areas|enums|stats_size|field_values|field_sizes` for API introspection and field stats.\r\n- If the user needs full pages or aggregated responses, set `save_raw=true` and report the saved file path.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required field: `action`\r\n- Supported actions: `studies`, `metadata`, `search_areas`, `enums`, `stats_size`, `field_values`, `field_sizes`, `request`\r\n- Optional fields: `path` for `action=request`, `params`, `max_items`, `max_depth`, `max_pages`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common ClinicalTrials.gov patterns:\r\n  - `{\"action\":\"studies\",\"params\":{\"query.cond\":\"prostate cancer\",\"filter.overallStatus\":\"RECRUITING\",\"pageSize\":10},\"max_items\":10,\"max_pages\":1}`\r\n  - `{\"action\":\"metadata\"}`\r\n  - `{\"action\":\"field_values\",\"params\":{\"field\":\"protocolSection.identificationModule.organization.fullName\"}}`\r\n\r\n## Output\r\n- `action=studies` returns `pages_fetched`, `next_page_token`, count metadata, and compact `records`.\r\n- Other actions return either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"action\":\"studies\",\"params\":{\"query.cond\":\"prostate cancer\",\"filter.overallStatus\":\"RECRUITING\",\"pageSize\":10},\"max_items\":10,\"max_pages\":1}' | python scripts/clinicaltrials_client.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/clinicaltrials_client.py`.\n\nFile v1.0.3:skills/clinvar-variation-skill/SKILL.md\n\n---\r\nname: clinvar-variation-skill\r\ndescription: Submit compact ClinVar Clinical Tables and NCBI Variation requests for search, VCV, RCV, SCV, and RefSNP lookups. Use when a user wants variant-level summaries or identifier mapping\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/clinvar_variation.py` for all ClinVar and NCBI Variation work.\r\n- The script accepts `max_items`; for `action=search`, start around `max_items=10`.\r\n- For `vcv`, `rcv`, `scv`, and `refsnp`, omit `max_items` unless you need to trim nested arrays in the summary.\r\n- Re-run requests in long conversations instead of relying on prior tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\r\n- If the user asks for full JSON, set `save_raw=true` and report the saved file path instead of pasting large payloads into chat.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return the JSON verbatim only if the user explicitly asks for machine-readable output.\r\n- Use `action=search` for the Clinical Tables endpoint.\r\n- Use `action=vcv|rcv|scv|refsnp` for NCBI Variation beta objects.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required field: `action`\r\n- Action-specific required fields:\r\n  - `search`: `terms`\r\n  - `vcv`: `vcv`\r\n  - `rcv`: `rcv`\r\n  - `scv`: `scv`\r\n  - `refsnp`: `refsnp`\r\n- Optional fields: `params`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n\r\n## Output\r\n- `search` returns `total`, `identifiers`, `display_rows`, `extra_fields`, and truncation metadata.\r\n- `vcv|rcv|scv|refsnp` return a compact `summary` and optional `top_keys`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failures return `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"action\":\"search\",\"terms\":\"VCV000013080\",\"max_items\":10}' | python scripts/clinvar_variation.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/clinvar_variation.py`.\n\nFile v1.0.3:skills/efo-ontology-skill/SKILL.md\n\n---\r\nname: efo-ontology-skill\r\ndescription: Submit compact EFO OLS4 requests for search, term lookup, children, and descendants. Use when a user wants concise EFO resolution or ontology-expansion summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/rest_request.py` for all OLS4 and EFO API calls.\r\n- Use `base_url=https://www.ebi.ac.uk/ols4/api`.\r\n- Search, children, and descendant endpoints are better with `max_items=10`; single term lookups usually do not need `max_items`.\r\n- Use the smallest ontology expansion that answers the question.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Prefer these paths: `search`, `ontologies/efo/terms/<double-encoded-iri>`, and the corresponding `children` or `descendants` paths.\r\n- If the user needs the full payload, set `save_raw=true` and report the saved file path.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required fields: `base_url`, `path`\r\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common OLS4 patterns:\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/ols4/api\",\"path\":\"search\",\"params\":{\"q\":\"asthma\",\"ontology\":\"efo\"},\"record_path\":\"response.docs\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/ols4/api\",\"path\":\"ontologies/efo/terms/http%253A%252F%252Fwww.ebi.ac.uk%252Fefo%252FEFO_0000270\"}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/ols4/api\",\"path\":\"ontologies/efo/terms/http%253A%252F%252Fwww.ebi.ac.uk%252Fefo%252FEFO_0000270/descendants\",\"record_path\":\"_embedded.terms\",\"max_items\":10}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"base_url\":\"https://www.ebi.ac.uk/ols4/api\",\"path\":\"search\",\"params\":{\"q\":\"asthma\",\"ontology\":\"efo\"},\"record_path\":\"response.docs\",\"max_items\":10}' | python scripts/rest_request.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.3:skills/ensembl-skill/SKILL.md\n\n---\r\nname: ensembl-skill\r\ndescription: Submit compact Ensembl REST API requests for lookup, overlap, cross-reference, and variation endpoints. Use when a user wants concise Ensembl summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/rest_request.py` for all Ensembl API calls.\r\n- Use `base_url=https://rest.ensembl.org`.\r\n- The script accepts `max_items`; object lookups usually do not need it, but `overlap` and `xrefs` are better with `max_items=10`.\r\n- Send JSON-friendly headers such as `Accept: application/json` and `Content-Type: application/json`.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not part of the true request.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return the script JSON verbatim only if the user explicitly asks for machine-readable output.\r\n- Prefer these paths: `lookup/id/<id>`, `overlap/region/<species>/<region>`, `xrefs/id/<id>`, and `variation/<species>/<id>`.\r\n- Use `save_raw=true` when the user needs the full payload.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required fields: `base_url`, `path`\r\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common Ensembl patterns:\r\n  - `{\"base_url\":\"https://rest.ensembl.org\",\"path\":\"lookup/id/ENSG00000141510\",\"headers\":{\"Accept\":\"application/json\",\"Content-Type\":\"application/json\"}}`\r\n  - `{\"base_url\":\"https://rest.ensembl.org\",\"path\":\"overlap/region/homo_sapiens/1:1000000-1002000\",\"params\":{\"feature\":\"gene\"},\"headers\":{\"Accept\":\"application/json\",\"Content-Type\":\"application/json\"},\"max_items\":10}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"base_url\":\"https://rest.ensembl.org\",\"path\":\"lookup/id/ENSG00000141510\",\"headers\":{\"Accept\":\"application/json\",\"Content-Type\":\"application/json\"}}' | python scripts/rest_request.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.3:skills/gnomad-graphql-skill/SKILL.md\n\n---\r\nname: gnomad-graphql-skill\r\ndescription: Submit compact gnomAD GraphQL requests for frequency, gene constraint, and variant context queries. Use when a user wants concise gnomAD summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/gnomad_graphql.py` for all gnomAD GraphQL work.\r\n- For nested GraphQL results, start with `max_items=3` to `5`.\r\n- Keep selection sets narrow and page or filter at the query level instead of asking for broad dumps.\r\n- Use `query_path` for long GraphQL documents instead of pasting large inline queries.\r\n- Re-run requests in long conversations instead of relying on earlier tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not part of the real query.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return raw JSON only if the user explicitly asks for machine-readable output.\r\n- Prefer targeted queries for variant frequency, gene constraint, or transcript consequence context.\r\n- If the user needs the full payload, set `save_raw=true` and report the saved file path.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required field: `query` or `query_path`\r\n- Optional fields: `variables`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common gnomAD patterns:\r\n  - `{\"query\":\"query { meta { clinvar_release_date } }\"}`\r\n  - `{\"query\":\"query Variant($variantId: String!, $dataset: DatasetId!) { variant(variantId: $variantId, dataset: $dataset) { variantId genome { ac an af } } }\",\"variables\":{\"variantId\":\"1-55516888-G-GA\",\"dataset\":\"gnomad_r4\"},\"max_items\":3}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `top_keys`, a compact `summary`, and `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` such as `invalid_json`, `invalid_input`, `network_error`, `invalid_response`, or `graphql_error`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"query\":\"query { meta { clinvar_release_date } }\"}' | python scripts/gnomad_graphql.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/gnomad_graphql.py`.\n\nFile v1.0.3:skills/gwas-catalog-skill/SKILL.md\n\n---\r\nname: gwas-catalog-skill\r\ndescription: Submit compact GWAS Catalog REST API v2 requests for studies, associations, SNPs, EFO traits, genes, publications, loci, and metadata. Use when a user wants concise GWAS Catalog summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/rest_request.py` for all GWAS Catalog API calls.\r\n- Use `base_url=https://www.ebi.ac.uk/gwas/rest/api/v2`.\r\n- The script accepts `max_items`; for collection endpoints, start with API `size=10` and `max_items=10`.\r\n- Single-resource endpoints such as `studies/<accession>` generally do not need `max_items`.\r\n- Use `record_path` to target `_embedded.<resource>` lists.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return the script JSON verbatim only if the user explicitly asks for machine-readable output.\r\n- Prefer these paths: `metadata`, `studies`, `studies/<accession>`, `associations`, `snps`, `efoTraits`, `genes`, `publications`, and `loci`.\r\n- Use `save_raw=true` if the user needs the full HATEOAS payload or pagination links.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required fields: `base_url`, `path`\r\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common GWAS Catalog patterns:\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/gwas/rest/api/v2\",\"path\":\"metadata\"}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/gwas/rest/api/v2\",\"path\":\"studies\",\"params\":{\"efo_trait\":\"asthma\",\"size\":10},\"record_path\":\"_embedded.studies\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/gwas/rest/api/v2\",\"path\":\"associations\",\"params\":{\"mapped_gene\":\"BRCA1\",\"size\":10},\"record_path\":\"_embedded.associations\",\"max_items\":10}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"base_url\":\"https://www.ebi.ac.uk/gwas/rest/api/v2\",\"path\":\"studies\",\"params\":{\"efo_trait\":\"asthma\",\"size\":10},\"record_path\":\"_embedded.studies\",\"max_items\":10}' | python scripts/rest_request.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.3:skills/ncbi-clinicaltables-skill/SKILL.md\n\n---\r\nname: ncbi-clinicaltables-skill\r\ndescription: Submit compact Clinical Tables NCBI Gene requests for human gene lookup, pagination, and field selection. Use when a user wants concise autocomplete-style human gene search results\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/ncbi_gene_clinicaltables.py` for all Clinical Tables gene searches.\r\n- The script accepts `max_items`; for search pages, start with `count=10` and `max_items=10`.\r\n- Use `params` for endpoint options like `df`, `ef`, `sf`, `q`, `offset`, and `count`.\r\n- Prefer `ncbi-entrez-skill` when the user wants general Entrez Gene records rather than autocomplete/search rows.\r\n- Page with `offset` instead of asking for large pulls.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\r\n- If the user asks for the full payload, set `save_raw=true` and report the saved file path instead of pasting large response arrays into chat.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return the JSON verbatim only if the user explicitly asks for machine-readable output.\r\n- Use `terms` for the primary search text.\r\n- Keep `count` modest and page with `offset` instead of pulling large result sets at once.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required field: `terms`\r\n- Optional fields: `params`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common NCBI Gene patterns:\r\n  - `{\"terms\":\"TP53\",\"params\":{\"df\":\"GeneID,Symbol,description\"}}`\r\n  - `{\"terms\":\"BRCA\",\"params\":{\"count\":10,\"df\":\"chromosome,GeneID,Symbol,description,type_of_gene\"},\"max_items\":10}`\r\n  - `{\"terms\":\"kinase\",\"params\":{\"count\":10,\"offset\":10,\"df\":\"GeneID,Symbol,description\"},\"max_items\":10}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `terms`, `total`, `codes`, `display_rows`, `extra_fields`, and truncation metadata.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"terms\":\"TP53\",\"params\":{\"count\":10,\"df\":\"GeneID,Symbol,description\"},\"max_items\":10}' | python scripts/ncbi_gene_clinicaltables.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/ncbi_gene_clinicaltables.py`.\n\nFile v1.0.3:skills/ncbi-entrez-skill/SKILL.md\n\n---\r\nname: ncbi-entrez-skill\r\ndescription: Submit compact NCBI Entrez E-Utilities requests for PubMed, Gene, Protein, Nucleotide, PMC metadata, and GEO metadata workflows. Use when a user wants concise Entrez search, fetch, summary, or link results; save raw JSON or XML only on request.\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/ncbi_entrez.py` for all Entrez calls in this package.\r\n- Use explicit `endpoint` values such as `esearch`, `esummary`, `efetch`, `elink`, or `einfo`.\r\n- Search-style Entrez calls are better with `retmax=10` and `max_items=10`.\r\n- GEO is nested under this skill. Use `db=gds` or `db=geoprofiles` for GEO metadata and load `references/geo.md` only when the user is specifically asking about GEO.\r\n- BLAST workflows belong in `ncbi-blast-skill`. PMC Open Access workflows belong in `ncbi-pmc-skill`. Datasets v2 workflows belong in `ncbi-datasets-skill`.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script output by default.\r\n- Return raw JSON or XML only if the user explicitly asks for machine-readable output.\r\n- Prefer targeted endpoint calls instead of broad unfiltered dumps.\r\n- If the user needs the full raw response, set `save_raw=true` and report the saved file path.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required field: `endpoint`\r\n- Optional fields: `params`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common Entrez patterns:\r\n  - `{\"endpoint\":\"esearch\",\"params\":{\"db\":\"pubmed\",\"term\":\"KRAS AND colorectal cancer\",\"retmode\":\"json\",\"retmax\":10},\"max_items\":10}`\r\n  - `{\"endpoint\":\"esummary\",\"params\":{\"db\":\"gene\",\"id\":\"7157\",\"retmode\":\"json\"},\"max_items\":10}`\r\n  - `{\"endpoint\":\"efetch\",\"params\":{\"db\":\"protein\",\"id\":\"NP_000537.3\",\"retmode\":\"xml\"},\"response_format\":\"xml\",\"max_items\":10}`\r\n  - `{\"endpoint\":\"elink\",\"params\":{\"dbfrom\":\"gds\",\"db\":\"pubmed\",\"id\":\"200000001\",\"retmode\":\"json\"},\"max_items\":10}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, endpoint metadata, and either compact `records`, a compact `summary`, or `text_head`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"endpoint\":\"esearch\",\"params\":{\"db\":\"gene\",\"term\":\"TP53[gene] AND human[orgn]\",\"retmode\":\"json\",\"retmax\":10},\"max_items\":10}' | python scripts/ncbi_entrez.py\r\n```\r\n\r\n## References\r\n- Load `references/geo.md` only when the user specifically needs GEO query patterns.\r\n- Keep the import package limited to this file, `references/geo.md`, and `scripts/ncbi_entrez.py`.\n\nArchive v1.0.2: 50 files, 111357 bytes\n\nFiles: references/drug-naming.md (4091b), references/pharma-intelligence-workflow.md (8687b), references/regulatory-timelines.md (5149b), references/sources-by-region.md (8513b), references/sub-skills.md (8729b), skill-card.md (2538b), SKILL.md (15002b), skills/biorxiv-skill/scripts/rest_request.py (10728b), skills/biorxiv-skill/SKILL.md (2515b), skills/chebi-skill/scripts/rest_request.py (10728b), skills/chebi-skill/SKILL.md (2178b), skills/chembl-skill/scripts/rest_request.py (10728b), skills/chembl-skill/SKILL.md (2465b), skills/clinicaltrials-skill/scripts/clinicaltrials_client.py (8796b), skills/clinicaltrials-skill/SKILL.md (2386b), skills/clinvar-variation-skill/scripts/clinvar_variation.py (7835b), skills/clinvar-variation-skill/SKILL.md (2007b), skills/efo-ontology-skill/scripts/rest_request.py (10728b), skills/efo-ontology-skill/SKILL.md (2411b), skills/ensembl-skill/scripts/rest_request.py (10728b), skills/ensembl-skill/SKILL.md (2368b), skills/gnomad-graphql-skill/scripts/gnomad_graphql.py (5616b), skills/gnomad-graphql-skill/SKILL.md (2109b), skills/gwas-catalog-skill/scripts/rest_request.py (10728b), skills/gwas-catalog-skill/SKILL.md (2564b), skills/ncbi-clinicaltables-skill/scripts/ncbi_gene_clinicaltables.py (5559b), skills/ncbi-clinicaltables-skill/SKILL.md (2368b), skills/ncbi-entrez-skill/references/geo.md (930b), skills/ncbi-entrez-skill/scripts/ncbi_entrez.py (10830b), skills/ncbi-entrez-skill/SKILL.md (2730b), skills/ncbi-pmc-skill/scripts/ncbi_pmc.py (8755b), skills/ncbi-pmc-skill/SKILL.md (1808b), skills/opentargets-skill/scripts/opentargets_disease_heatmap.py (12104b), skills/opentargets-skill/scripts/opentargets_graphql.py (5646b), skills/opentargets-skill/SKILL.md (3238b), skills/pubchem-pug-skill/scripts/rest_request.py (10728b), skills/pubchem-pug-skill/SKILL.md (2367b), skills/quickgo-skill/scripts/rest_request.py (10728b), skills/quickgo-skill/SKILL.md (2458b), skills/reactome-skill/scripts/rest_request.py (10728b), skills/reactome-skill/SKILL.md (2396b), skills/rhea-skill/scripts/rest_request.py (10728b), skills/rhea-skill/SKILL.md (1990b), skills/rnacentral-skill/scripts/rest_request.py (10728b), skills/rnacentral-skill/SKILL.md (2209b), skills/string-skill/scripts/rest_request.py (10728b), skills/string-skill/SKILL.md (2555b), skills/uniprot-skill/scripts/rest_request.py (10728b), skills/uniprot-skill/SKILL.md (2733b), _meta.json (138b)\n\nFile v1.0.2:SKILL.md\n\n---\nname: pharma-intelligence\ndescription: >\n  In-depth, multi-region pharmaceutical intelligence search and synthesis,\n  plus drug repurposing, target discovery, clinical evidence review, and\n  bioactivity analysis. Use this skill whenever the user asks about drug\n  approvals, clinical trials, regulatory submissions, pipeline assets, patent\n  landscapes, competitive intelligence, scientific evidence, disease targets,\n  genetic associations, or compound bioactivity for any drug, target,\n  indication, or company — especially when coverage of China, US, Europe,\n  Japan, South Korea, or Australia is needed. Trigger even for casual queries\n  like \"what's the approval status of X in China\", \"find trials for Y in\n  Japan\", \"compare pipeline coverage across regions\", \"find drugs for disease\n  Z\", or \"what targets are associated with condition W\". Always consult this\n  skill before answering any pharma or biomedical research question that\n  requires source-grounded data.\n---\n\n# Global Pharma Intelligence & Biomedical Research Skill\n\nSystematic, source-prioritized search and synthesis across regulatory, clinical,\nacademic, and commercial databases — covering all major pharmaceutical markets\nand 14+ biomedical research databases.\n\n## Sub-Skills — How to Invoke\n\nThis skill delegates all database work to the sub-skills bundled locally under `skills/`.\nRead the relevant sub-skill's `SKILL.md` before invoking it, then run its bundled script.\n\nSee [references/sub-skills.md](./references/sub-skills.md) for the full mapping of research\ntasks to sub-skills and execution patterns.\n\n---\n\n## Core Principle: Tiered Source Priority\n\nEvery region follows a 3-tier hierarchy. Higher tiers override lower-tier claims; always cite the tier.\n\n| Tier | Type | Description |\n|------|------|-------------|\n| **Tier 1** | Regulatory | Official agency submissions, approvals, labels |\n| **Tier 2** | Trial registries | Prospective/registered clinical evidence |\n| **Tier 3** | Academic / IP | Published papers, conferences, patents |\n\nFor the per-region source map (CN / US / EU / JP / KR / AU + global) with URLs and access notes, see [references/sources-by-region.md](./references/sources-by-region.md).\n\n---\n\n## Search Workflow\n\n### Step 1 — Classify the Query (pick ONE intent)\n\n| # | Intent | Trigger phrases |\n|---|--------|-----------------|\n| A | Trial landscape | \"trials of X\", \"clinical studies of\", \"who is testing\", \"phase 2/3 of\" |\n| B | Approval / regulatory status | \"is X approved\", \"approval status\", \"FDA/EMA/NMPA cleared\" |\n| C | Safety / adverse events | \"side effects of\", \"is X safe\", \"adverse events\", \"black box\" |\n| D | Pipeline / competitive intel | \"pipeline\", \"competitive landscape\", \"who else is developing\" |\n| E | Patent / IP / exclusivity | \"when does patent expire\", \"patent landscape\", \"exclusivity\" |\n| F | Target / mechanism / drug discovery | \"drugs targeting X\", \"mechanism of\", \"bioactivity\", \"IC50\" |\n| G | Repurposing / target discovery | \"repurpose for\", \"targets associated with disease\", \"genetic basis\" |\n| H | Literature / evidence review | \"recent papers on\", \"what's known about\", \"systematic review\" |\n\nAlso capture: **regions in scope** (US / EU / JP / CN / KR / AU / global) and **time horizon**.\n\n### Step 2 — Execute the Per-Intent Sequence\n\nRun the workflow for the chosen intent (see [Per-Intent Workflows](#per-intent-workflows)) in order. For sources without MCP coverage (CN NMPA/CDE, EMA EPAR, PMDA, jRCT, CTIS, CRIS, ANZCTR, Orange Book), use `web_fetch` only at the steps that name them.\n\nResolve identifiers as needed:\n- Free-text disease → MONDO/EFO ID via `opentargets-skill` or `efo-ontology-skill`\n- Free-text gene → HGNC symbol via `ncbi-clinicaltables-skill` or `ensembl-skill`\n- Cross-database ID conversion → `ensembl-skill`, `uniprot-skill`, or `efo-ontology-skill`\n\n### Step 3 — Resolve Conflicts\n\n1. Higher-tier source wins (Tier 1 > Tier 2 > Tier 3).\n2. More recent data wins within the same tier.\n3. Flag unresolved conflicts; do not silently pick one.\n\n### Step 4 — Synthesize and Present\n\nStructure output to match the intent of the question:\n- Trial landscape → table of trials (NCT/registry ID, phase, status, sponsor, N, primary endpoint).\n- Approval status → region × status × date × indications table.\n- Safety → top FAERS reactions plus black-box / warnings.\n- Pipeline → drug × company × phase × mechanism table.\n- Patent → patent number, jurisdiction, expiry.\n\nAlways cite source, tier, and access date.\n\n---\n\n## Per-Intent Workflows\n\n### A. Trial Landscape\n\n*\"What clinical studies / trials exist for [drug | target | indication]?\"*\n\nDefault scope = ALL regions. Only narrow if the user names a single region.\n\n   `clinicaltrials-skill` covers only ClinicalTrials.gov, which is primarily US-registered trials. Run each regional source in parallel.\n\n1. **United States** — `clinicaltrials-skill` (`action=studies`).\n   - Use `query.intr` for a drug, `query.cond` for a disease, both for combined.\n   - For a target/class (e.g., \"pan-RAS\", \"PD-L1 inhibitor\"): pass the class term as `query.intr` plus a relevant `query.cond`.\n   - Then re-run with an NCT ID in `query.id` for eligibility, endpoints, sponsor, and locations.\n2. **China** — `web_fetch`:\n   - `http://www.chinadrugtrials.org.cn` (mandatory CN IND registry)\n   - `https://www.chictr.org.cn` (ChiCTR, WHO primary)\n3. **Europe** — `web_fetch`:\n   - `https://euclinicaltrials.eu` (CTIS — current EU register)\n   - `https://eudract.ema.europa.eu` (EudraCT — legacy historical trials)\n   - `https://www.isrctn.com` (ISRCTN, UK/global)\n4. **Japan** — `web_fetch`:\n   - `https://jrct.niph.go.jp` (jRCT — mandatory JP registry)\n   - `https://www.umin.ac.jp/ctr/` (UMIN-CTR — legacy)\n5. **South Korea** — `web_fetch` `https://cris.nih.go.kr`.\n6. **Australia / New Zealand** — `web_fetch` `https://www.anzctr.org.au`.\n7. **WHO ICTRP catch-all** — `web_fetch` `https://trialsearch.who.int` for any WHO primary registry (covers India CTRI, Iran IRCT, Brazil ReBEC, etc.).\n8. **Published results** — `ncbi-entrez-skill` (`db=pubmed`) with NCT ID or drug name to surface completed-trial papers.\n9. **US company-disclosed pipeline** (optional) — `web_fetch` SEC EDGAR full-text search at `https://efts.sec.gov/LATEST/search-index` for US-listed sponsors.\n\nFor every regional `web_fetch`: query both INN and brand name; for CN also use the Chinese transliteration (see [references/drug-naming.md](./references/drug-naming.md)). Aggregate results in one table with a \"Registry\" column.\n\n### B. Approval / Regulatory Status\n\n*\"Is [drug] approved in [region]?\"*\n\n1. **US** — `web_fetch` `https://api.fda.gov/drug/label.json` (openFDA) and `https://dailymed.nlm.nih.gov/dailymed/services/v2/spls.json` (label date anchors approval); `web_fetch` `https://api.fda.gov/drug/ndc.json` for orphan status.\n2. **Non-US** — `web_fetch` the regional Tier 1 source (NMPA, EMA EPAR, PMDA, MFDS, TGA). For CN, also search Chinese characters.\n3. `chembl-skill` (`drug_indication.json?molecule_chembl_id=...`) — cross-check approved indications and max phase.\n4. Say \"not approved\" only when Tier 1 affirms denial/withdrawal. Otherwise: \"no record found as of [date]\".\n\n### C. Safety / Adverse Events\n\n1. `web_fetch` `https://api.fda.gov/drug/event.json` (FAERS) filtering by drug name and seriousness.\n2. `web_fetch` `https://api.fda.gov/drug/label.json` with `sections=warnings` and `sections=contraindications`.\n3. `chembl-skill` (`molecule/<id>.json`) — inspect the `black_box_warning` flag.\n4. `ncbi-entrez-skill` (`db=pubmed`) with terms `\"adverse effect\" OR \"toxicity\"` for case reports and post-marketing literature.\n\n### D. Pipeline / Competitive Intelligence\n\n*\"Who else is developing for [indication / target]? What's the global competitive landscape?\"*\n\nDefault scope = ALL regions. A competitive landscape without the active-trial picture is incomplete, so run the full multi-region trial sweep from Workflow A and then layer pipeline-specific sources on top.\n\n1. **Active trials — all regions** — run [Workflow A](#a-trial-landscape) steps 1–7 in full, optionally adding `filter.overallStatus=RECRUITING` (or `ACTIVE_NOT_RECRUITING`) and a phase filter to focus on competitors at a specific stage.\n2. **Company disclosures** — `web_fetch` SEC EDGAR full-text search at `https://efts.sec.gov/LATEST/search-index` for pipeline language in 10-K / 10-Q / 8-K (US-listed sponsors only).\n3. **Patent activity per company** — `web_fetch` `https://patents.google.com` with an `assignee:` filter (or WIPO PATENTSCOPE / Espacenet — see Workflow E).\n4. **Published results** — `ncbi-entrez-skill` (`db=pubmed`) with NCT IDs or drug names to surface completed-trial papers.\n\nAggregate into one table: drug × company × phase × mechanism × registry/region.\n\n### E. Patent / IP / Exclusivity\n\nAll listed patent sources are free and require no API key.\n\n1. **Global patent search** — `web_fetch` one or more of:\n   - `https://patents.google.com` (Google Patents — best full-text search, covers USPTO, EPO, WIPO, JPO, CNIPA, KIPO).\n   - `https://patentscope.wipo.int` (WIPO PATENTSCOPE — authoritative for PCT applications and national filings worldwide).\n   - `https://worldwide.espacenet.com` (EPO Espacenet — strongest European and family-tree coverage).\n2. **US patents (structured)** — `uspto_ppubs_search_patents` via MCP for granted patents and applications.\n3. **Patent family / cross-jurisdiction equivalents** — Espacenet's \"INPADOC patent family\" view, or Google Patents' \"Worldwide applications\" section.\n4. **Orange Book** (patent + exclusivity expiry for FDA-approved drugs) — `web_fetch` `https://www.accessdata.fda.gov/scripts/cder/ob`.\n5. **Orphan exclusivity** — `fda_orphan_search_exclusivity` (7-year US orphan exclusivity).\n\n### F. Target / Mechanism / Drug Discovery\n\n1. `chembl-skill` — search `target/search.json?q=<gene>` to resolve target ChEMBL ID, then `mechanism.json?target_chembl_id=...` for all drugs.\n2. `chembl-skill` — `mechanism.json?molecule_chembl_id=...` for mechanism of action of each candidate.\n3. `chembl-skill` — `activity.json?target_chembl_id=...` for IC50 / Kd / EC50 bioactivity comparisons.\n4. `uniprot-skill` — `uniprotkb/search` with `gene:<symbol> AND organism_id:9606` for protein function and druggability context.\n5. `reactome-skill` — pathway and disease-pathway context for the target.\n\n### G. Repurposing / Target Discovery\n\n1. `opentargets-skill` — search for disease to resolve MONDO / EFO ID.\n2. `opentargets-skill` — `associatedTargets` query with disease EFO ID → ranked targets by evidence score.\n3. `gwas-catalog-skill` — associations for the disease EFO term to identify genetically supported targets.\n4. `web_fetch` OMIM API at `https://api.omim.org/api/entry/search` for Mendelian basis (requires API key).\n5. For each top target: `chembl-skill` — `mechanism.json?target_chembl_id=...` for all drugs.\n6. `clinicaltrials-skill` with each drug as `query.intr` for prior-art trials.\n7. `web_fetch` `https://api.fda.gov/drug/event.json` as a safety filter for non-trivial candidates.\n\n### H. Literature / Evidence Review\n\n1. `ncbi-entrez-skill` (`db=pubmed`) — entry point; use MeSH terms for disease, chemical, and gene-aware filtering.\n2. `web_fetch` Europe PMC REST (`https://www.ebi.ac.uk/europepmc/webservices/rest/search`) — broader: grants, preprints, non-MEDLINE.\n3. `biorxiv-skill` — bioRxiv / medRxiv preprints only.\n4. `ncbi-pmc-skill` or `ncbi-entrez-skill` (`efetch`, `db=pmc`) — abstract or full text for top hits.\n\n---\n\n## Combination Strategies (cross-intent)\n\nUse only when a question genuinely spans multiple intents.\n\n- **Disease → Targets → Drugs → Trials**: `opentargets-skill` (search + associations) → `chembl-skill` (mechanism by target) → `clinicaltrials-skill`\n- **Gene → Protein → Pathways → Drugs**: `ncbi-clinicaltables-skill` or `ensembl-skill` → `uniprot-skill` → `reactome-skill` → `chembl-skill` (mechanism by target)\n- **Variant → Gene → Disease → Treatments**: `clinvar-variation-skill` or `gnomad-graphql-skill` → `ensembl-skill` → `opentargets-skill` → `chembl-skill` (mechanism by target)\n- **Drug → Safety → Label → Trials**: `chembl-skill` (mechanism) → `web_fetch` openFDA adverse events → `web_fetch` openFDA label → `clinicaltrials-skill`\n\n---\n\n## API Keys\n\nMost APIs require no key. Exceptions:\n\n| Database | Key | Source |\n|----------|-----|--------|\n| OMIM | Required | https://omim.org/api |\n| NCI Clinical Trials | Optional | https://clinicaltrialsapi.cancer.gov |\n| OpenFDA | Optional (higher rate limits) | https://open.fda.gov/apis |\n\nAll bundled sub-skills (ChEMBL, OpenTargets, PubMed via NCBI Entrez, ClinicalTrials.gov, Reactome, UniProt, GWAS Catalog, Ensembl) are public and require no key. Patent landscape work uses Google Patents, WIPO PATENTSCOPE, and Espacenet via `web_fetch` — no keys required.\n\n---\n\n## Output Quality Standards\n\n- Never fabricate approval dates, trial IDs, or efficacy numbers.\n- Attribute every claim to its source and tier.\n- Flag gaps explicitly (e.g., \"No registered trials found in jRCT as of [date]\").\n- Distinguish \"no data found\" from \"not approved\" — absence of evidence ≠ negative regulatory decision.\n- For Chinese sources: note whether the search was conducted in Chinese characters; romanization alone may miss records.\n\n---\n\n## Troubleshooting\n\n**No results?**\n- Try alternative terms (INN vs brand name, gene symbol vs protein name).\n- Use standardized IDs: MONDO/EFO for diseases, HGNC for genes, ChEMBL IDs for compounds, Ensembl for OpenTargets.\n- Resolve IDs first with `efo-ontology-skill`, `ncbi-clinicaltables-skill`, `ensembl-skill`, or `uniprot-skill`.\n\n**Too many results?**\n- Add filters: `max_items`, `filter.phase`, `filter.overallStatus`, `reviewed=true` (UniProt).\n- Apply date ranges where supported.\n\n**API key errors?**\n- OMIM requires a key; NCI and OpenFDA accept optional keys for higher rate limits.\n\n**Source not covered by a sub-skill?**\n- Use `web_fetch` directly for CDE/NMPA, EMA/EPAR, PMDA, jRCT, CTIS, CRIS, ANZCTR, Orange Book, openFDA, DailyMed, FAERS, and EDGAR.\n\n---\n\n## References\n\n- [references/sub-skills.md](./references/sub-skills.md) — Mapping of pharma-intelligence tasks to bundled sub-skills, with execution patterns.\n- [references/drug-naming.md](./references/drug-naming.md) — INN / brand / Chinese / Japanese naming conventions and transliteration.\n- [references/regulatory-timelines.md](./references/regulatory-timelines.md) — Review-clock lengths and milestones per agency (FDA, EMA, PMDA, CDE/NMPA, etc.).\n- [references/sources-by-region.md](./references/sources-by-region.md) — Direct URLs and access notes for all regional regulatory databases.\n- [references/pharma-intelligence-workflow.md](./references/pharma-intelligence-workflow.md) — End-to-end worked example (osimertinib in NSCLC).\n\nFile v1.0.2:skills/biorxiv-skill/SKILL.md\n\n---\nname: biorxiv-skill\ndescription: Submit compact bioRxiv and medRxiv API requests for details, publication-linkage, and DOI lookups. Use when a user wants concise preprint metadata summaries\n---\n\n## Operating rules\n- Use `scripts/rest_request.py` for all bioRxiv and medRxiv API calls.\n- Use `base_url=https://api.biorxiv.org`.\n- The script accepts `max_items`; for `details` and `pubs` pages, start around `max_items=10`.\n- Prefer one cursor page at a time instead of increasing page size or pasting long collections into chat.\n- Re-run requests in long conversations instead of relying on older tool output.\n- Treat displayed `...` in tool previews as UI truncation, not part of the true request.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Return the raw script JSON only if the user explicitly asks for machine-readable output.\n- Prefer these paths: `details/<server>/<start>/<end>/<cursor>/json`, `details/<server>/<doi>/na/json`, `pubs/<server>/<start>/<end>/<cursor>`, and `pubs/<server>/<doi>/na/json`.\n- If the user needs full page contents, set `save_raw=true` and report the saved file path rather than pasting large collections into chat.\n\n## Input\n- Read one JSON object from stdin.\n- Required fields: `base_url`, `path`\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common biorxiv patterns:\n  - `{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/biorxiv/2025-03-21/2025-03-28/0/json\",\"record_path\":\"collection\",\"max_items\":10}`\n  - `{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/medrxiv/10.1101/2020.09.09.20191205/na/json\",\"record_path\":\"collection\",\"max_items\":10}`\n  - `{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"pubs/medrxiv/2020-03-01/2020-03-30/0\",\"record_path\":\"collection\",\"max_items\":10}`\n\n## Output\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\n- Use `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/biorxiv/2025-03-21/2025-03-28/0/json\",\"record_path\":\"collection\",\"max_items\":10}' | python scripts/rest_request.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.2:skills/chebi-skill/SKILL.md\n\n---\nname: chebi-skill\ndescription: Submit compact ChEBI 2.0 API requests for chemical search, compound lookup, ontology traversal, and structure metadata. Use when a user wants concise ChEBI summaries\n---\n\n## Operating rules\n- Use `scripts/rest_request.py` for all ChEBI calls.\n- Use `base_url=https://www.ebi.ac.uk`.\n- Prefer the documented public routes under `chebi/backend/api/public/`.\n- Start with `es_search/` for free-text lookup and use `compound/<CHEBI:id>/` for targeted records.\n- Re-run requests in long conversations instead of relying on older tool output.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Return raw JSON only if the user explicitly asks for machine-readable output.\n- Prefer these paths: `chebi/backend/api/public/es_search/`, `chebi/backend/api/public/compound/<CHEBI:id>/`, and ontology child or parent routes.\n\n## Input\n- Read one JSON object from stdin.\n- Required fields: `base_url`, `path`\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common ChEBI patterns:\n  - `{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/es_search/\",\"params\":{\"query\":\"caffeine\",\"size\":10},\"record_path\":\"results\",\"max_items\":10}`\n  - `{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/compound/CHEBI:27732/\"}`\n  - `{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/ontology/children/CHEBI:27732/\"}`\n\n## Output\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\n- Use `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/es_search/\",\"params\":{\"query\":\"caffeine\",\"size\":10},\"record_path\":\"results\",\"max_items\":10}' | python scripts/rest_request.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.2:skills/chembl-skill/SKILL.md\n\n---\nname: chembl-skill\ndescription: Submit compact ChEMBL API requests for activity, molecule, target, mechanism, and text-search endpoints. Use when a user wants concise ChEMBL summaries\n---\n\n## Operating rules\n- Use `scripts/rest_request.py` for all ChEMBL API calls.\n- Use `base_url=https://www.ebi.ac.uk/chembl/api/data`.\n- The script accepts `max_items`; for activity, mechanism, and text-search collections, start with API `limit=10` and `max_items=10`.\n- Single molecule or target lookups usually do not need `max_items`.\n- Re-run requests in long conversations instead of relying on older tool output.\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Return the script JSON verbatim only if the user explicitly asks for machine-readable output.\n- Prefer these paths: `activity.json`, `molecule/<id>.json`, `target/<id>.json`, `mechanism.json`, and `molecule/search.json`.\n- Use `record_path` to target list fields like `activities`, `mechanisms`, or `molecules`.\n\n## Input\n- Read one JSON object from stdin.\n- Required fields: `base_url`, `path`\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common ChEMBL patterns:\n  - `{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"activity.json\",\"params\":{\"molecule_chembl_id\":\"CHEMBL25\",\"limit\":10},\"record_path\":\"activities\",\"max_items\":10}`\n  - `{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"molecule/CHEMBL25.json\"}`\n  - `{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"molecule/search.json\",\"params\":{\"q\":\"imatinib\",\"limit\":10},\"record_path\":\"molecules\",\"max_items\":10}`\n\n## Output\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\n- Use `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"activity.json\",\"params\":{\"molecule_chembl_id\":\"CHEMBL25\",\"limit\":10},\"record_path\":\"activities\",\"max_items\":10}' | python scripts/rest_request.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.2:skills/clinicaltrials-skill/SKILL.md\n\n---\nname: clinicaltrials-skill\ndescription: Submit compact ClinicalTrials.gov API v2 requests for study search, metadata, enums, search areas, and field statistics. Use when a user wants concise ClinicalTrials.gov summaries\n---\n\n## Operating rules\n- Use `scripts/clinicaltrials_client.py` for all ClinicalTrials.gov v2 calls.\n- Study searches are better with `max_items=10` and `max_pages=1`; only increase pages when the user explicitly wants more than the first page.\n- Use targeted `params` instead of broad unfiltered study dumps.\n- Re-run requests in long conversations instead of relying on older tool output.\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Prefer `action=studies` for search and `action=metadata|search_areas|enums|stats_size|field_values|field_sizes` for API introspection and field stats.\n- If the user needs full pages or aggregated responses, set `save_raw=true` and report the saved file path.\n\n## Input\n- Read one JSON object from stdin.\n- Required field: `action`\n- Supported actions: `studies`, `metadata`, `search_areas`, `enums`, `stats_size`, `field_values`, `field_sizes`, `request`\n- Optional fields: `path` for `action=request`, `params`, `max_items`, `max_depth`, `max_pages`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common ClinicalTrials.gov patterns:\n  - `{\"action\":\"studies\",\"params\":{\"query.cond\":\"prostate cancer\",\"filter.overallStatus\":\"RECRUITING\",\"pageSize\":10},\"max_items\":10,\"max_pages\":1}`\n  - `{\"action\":\"metadata\"}`\n  - `{\"action\":\"field_values\",\"params\":{\"field\":\"protocolSection.identificationModule.organization.fullName\"}}`\n\n## Output\n- `action=studies` returns `pages_fetched`, `next_page_token`, count metadata, and compact `records`.\n- Other actions return either compact `records` or a compact `summary`.\n- Use `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"action\":\"studies\",\"params\":{\"query.cond\":\"prostate cancer\",\"filter.overallStatus\":\"RECRUITING\",\"pageSize\":10},\"max_items\":10,\"max_pages\":1}' | python scripts/clinicaltrials_client.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/clinicaltrials_client.py`.\n\nFile v1.0.2:skills/clinvar-variation-skill/SKILL.md\n\n---\nname: clinvar-variation-skill\ndescription: Submit compact ClinVar Clinical Tables and NCBI Variation requests for search, VCV, RCV, SCV, and RefSNP lookups. Use when a user wants variant-level summaries or identifier mapping\n---\n\n## Operating rules\n- Use `scripts/clinvar_variation.py` for all ClinVar and NCBI Variation work.\n- The script accepts `max_items`; for `action=search`, start around `max_items=10`.\n- For `vcv`, `rcv`, `scv`, and `refsnp`, omit `max_items` unless you need to trim nested arrays in the summary.\n- Re-run requests in long conversations instead of relying on prior tool output.\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\n- If the user asks for full JSON, set `save_raw=true` and report the saved file path instead of pasting large payloads into chat.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Return the JSON verbatim only if the user explicitly asks for machine-readable output.\n- Use `action=search` for the Clinical Tables endpoint.\n- Use `action=vcv|rcv|scv|refsnp` for NCBI Variation beta objects.\n\n## Input\n- Read one JSON object from stdin.\n- Required field: `action`\n- Action-specific required fields:\n  - `search`: `terms`\n  - `vcv`: `vcv`\n  - `rcv`: `rcv`\n  - `scv`: `scv`\n  - `refsnp`: `refsnp`\n- Optional fields: `params`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n\n## Output\n- `search` returns `total`, `identifiers`, `display_rows`, `extra_fields`, and truncation metadata.\n- `vcv|rcv|scv|refsnp` return a compact `summary` and optional `top_keys`.\n- Use `raw_output_path` when `save_raw=true`.\n- Failures return `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"action\":\"search\",\"terms\":\"VCV000013080\",\"max_items\":10}' | python scripts/clinvar_variation.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/clinvar_variation.py`.\n\nFile v1.0.2:skills/efo-ontology-skill/SKILL.md\n\n---\nname: efo-ontology-skill\ndescription: Submit compact EFO OLS4 requests for search, term lookup, children, and descendants. Use when a user wants concise EFO resolution or ontology-expansion summaries\n---\n\n## Operating rules\n- Use `scripts/rest_request.py` for all OLS4 and EFO API calls.\n- Use `base_url=https://www.ebi.ac.uk/ols4/api`.\n- Search, children, and descendant endpoints are better with `max_items=10`; single term lookups usually do not need `max_items`.\n- Use the smallest ontology expansion that answers the question.\n- Re-run requests in long conversations instead of relying on older tool output.\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Prefer these paths: `search`, `ontologies/efo/terms/<double-encoded-iri>`, and the corresponding `children` or `descendants` paths.\n- If the user needs the full payload, set `save_raw=true` and report the saved file path.\n\n## Input\n- Read one JSON object from stdin.\n- Required fields: `base_url`, `path`\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common OLS4 patterns:\n  - `{\"base_url\":\"https://www.ebi.ac.uk/ols4/api\",\"path\":\"search\",\"params\":{\"q\":\"asthma\",\"ontology\":\"efo\"},\"record_path\":\"response.docs\",\"max_items\":10}`\n  - `{\"base_url\":\"https://www.ebi.ac.uk/ols4/api\",\"path\":\"ontologies/efo/terms/http%253A%252F%252Fwww.ebi.ac.uk%252Fefo%252FEFO_0000270\"}`\n  - `{\"base_url\":\"https://www.ebi.ac.uk/ols4/api\",\"path\":\"ontologies/efo/terms/http%253A%252F%252Fwww.ebi.ac.uk%252Fefo%252FEFO_0000270/descendants\",\"record_path\":\"_embedded.terms\",\"max_items\":10}`\n\n## Output\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\n- Use `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"base_url\":\"https://www.ebi.ac.uk/ols4/api\",\"path\":\"search\",\"params\":{\"q\":\"asthma\",\"ontology\":\"efo\"},\"record_path\":\"response.docs\",\"max_items\":10}' | python scripts/rest_request.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.2:skills/ensembl-skill/SKILL.md\n\n---\nname: ensembl-skill\ndescription: Submit compact Ensembl REST API requests for lookup, overlap, cross-reference, and variation endpoints. Use when a user wants concise Ensembl summaries\n---\n\n## Operating rules\n- Use `scripts/rest_request.py` for all Ensembl API calls.\n- Use `base_url=https://rest.ensembl.org`.\n- The script accepts `max_items`; object lookups usually do not need it, but `overlap` and `xrefs` are better with `max_items=10`.\n- Send JSON-friendly headers such as `Accept: application/json` and `Content-Type: application/json`.\n- Re-run requests in long conversations instead of relying on older tool output.\n- Treat displayed `...` in tool previews as UI truncation, not part of the true request.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Return the script JSON verbatim only if the user explicitly asks for machine-readable output.\n- Prefer these paths: `lookup/id/<id>`, `overlap/region/<species>/<region>`, `xrefs/id/<id>`, and `variation/<species>/<id>`.\n- Use `save_raw=true` when the user needs the full payload.\n\n## Input\n- Read one JSON object from stdin.\n- Required fields: `base_url`, `path`\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common Ensembl patterns:\n  - `{\"base_url\":\"https://rest.ensembl.org\",\"path\":\"lookup/id/ENSG00000141510\",\"headers\":{\"Accept\":\"application/json\",\"Content-Type\":\"application/json\"}}`\n  - `{\"base_url\":\"https://rest.ensembl.org\",\"path\":\"overlap/region/homo_sapiens/1:1000000-1002000\",\"params\":{\"feature\":\"gene\"},\"headers\":{\"Accept\":\"application/json\",\"Content-Type\":\"application/json\"},\"max_items\":10}`\n\n## Output\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\n- Use `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"base_url\":\"https://rest.ensembl.org\",\"path\":\"lookup/id/ENSG00000141510\",\"headers\":{\"Accept\":\"application/json\",\"Content-Type\":\"application/json\"}}' | python scripts/rest_request.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.2:skills/gnomad-graphql-skill/SKILL.md\n\n---\nname: gnomad-graphql-skill\ndescription: Submit compact gnomAD GraphQL requests for frequency, gene constraint, and variant context queries. Use when a user wants concise gnomAD summaries\n---\n\n## Operating rules\n- Use `scripts/gnomad_graphql.py` for all gnomAD GraphQL work.\n- For nested GraphQL results, start with `max_items=3` to `5`.\n- Keep selection sets narrow and page or filter at the query level instead of asking for broad dumps.\n- Use `query_path` for long GraphQL documents instead of pasting large inline queries.\n- Re-run requests in long conversations instead of relying on earlier tool output.\n- Treat displayed `...` in tool previews as UI truncation, not part of the real query.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Return raw JSON only if the user explicitly asks for machine-readable output.\n- Prefer targeted queries for variant frequency, gene constraint, or transcript consequence context.\n- If the user needs the full payload, set `save_raw=true` and report the saved file path.\n\n## Input\n- Read one JSON object from stdin.\n- Required field: `query` or `query_path`\n- Optional fields: `variables`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common gnomAD patterns:\n  - `{\"query\":\"query { meta { clinvar_release_date } }\"}`\n  - `{\"query\":\"query Variant($variantId: String!, $dataset: DatasetId!) { variant(variantId: $variantId, dataset: $dataset) { variantId genome { ac an af } } }\",\"variables\":{\"variantId\":\"1-55516888-G-GA\",\"dataset\":\"gnomad_r4\"},\"max_items\":3}`\n\n## Output\n- Success returns `ok`, `source`, `top_keys`, a compact `summary`, and `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` such as `invalid_json`, `invalid_input`, `network_error`, `invalid_response`, or `graphql_error`.\n\n## Execution\n```bash\necho '{\"query\":\"query { meta { clinvar_release_date } }\"}' | python scripts/gnomad_graphql.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/gnomad_graphql.py`.\n\nFile v1.0.2:skills/gwas-catalog-skill/SKILL.md\n\n---\nname: gwas-catalog-skill\ndescription: Submit compact GWAS Catalog REST API v2 requests for studies, associations, SNPs, EFO traits, genes, publications, loci, and metadata. Use when a user wants concise GWAS Catalog summaries\n---\n\n## Operating rules\n- Use `scripts/rest_request.py` for all GWAS Catalog API calls.\n- Use `base_url=https://www.ebi.ac.uk/gwas/rest/api/v2`.\n- The script accepts `max_items`; for collection endpoints, start with API `size=10` and `max_items=10`.\n- Single-resource endpoints such as `studies/<accession>` generally do not need `max_items`.\n- Use `record_path` to target `_embedded.<resource>` lists.\n- Re-run requests in long conversations instead of relying on older tool output.\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Return the script JSON verbatim only if the user explicitly asks for machine-readable output.\n- Prefer these paths: `metadata`, `studies`, `studies/<accession>`, `associations`, `snps`, `efoTraits`, `genes`, `publications`, and `loci`.\n- Use `save_raw=true` if the user needs the full HATEOAS payload or pagination links.\n\n## Input\n- Read one JSON object from stdin.\n- Required fields: `base_url`, `path`\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common GWAS Catalog patterns:\n  - `{\"base_url\":\"https://www.ebi.ac.uk/gwas/rest/api/v2\",\"path\":\"metadata\"}`\n  - `{\"base_url\":\"https://www.ebi.ac.uk/gwas/rest/api/v2\",\"path\":\"studies\",\"params\":{\"efo_trait\":\"asthma\",\"size\":10},\"record_path\":\"_embedded.studies\",\"max_items\":10}`\n  - `{\"base_url\":\"https://www.ebi.ac.uk/gwas/rest/api/v2\",\"path\":\"associations\",\"params\":{\"mapped_gene\":\"BRCA1\",\"size\":10},\"record_path\":\"_embedded.associations\",\"max_items\":10}`\n\n## Output\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\n- Use `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"base_url\":\"https://www.ebi.ac.uk/gwas/rest/api/v2\",\"path\":\"studies\",\"params\":{\"efo_trait\":\"asthma\",\"size\":10},\"record_path\":\"_embedded.studies\",\"max_items\":10}' | python scripts/rest_request.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`.\n\nFile v1.0.2:skills/ncbi-clinicaltables-skill/SKILL.md\n\n---\nname: ncbi-clinicaltables-skill\ndescription: Submit compact Clinical Tables NCBI Gene requests for human gene lookup, pagination, and field selection. Use when a user wants concise autocomplete-style human gene search results\n---\n\n## Operating rules\n- Use `scripts/ncbi_gene_clinicaltables.py` for all Clinical Tables gene searches.\n- The script accepts `max_items`; for search pages, start with `count=10` and `max_items=10`.\n- Use `params` for endpoint options like `df`, `ef`, `sf`, `q`, `offset`, and `count`.\n- Prefer `ncbi-entrez-skill` when the user wants general Entrez Gene records rather than autocomplete/search rows.\n- Page with `offset` instead of asking for large pulls.\n- Re-run requests in long conversations instead of relying on older tool output.\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\n- If the user asks for the full payload, set `save_raw=true` and report the saved file path instead of pasting large response arrays into chat.\n\n## Execution behavior\n- Return concise markdown summaries from the script JSON by default.\n- Return the JSON verbatim only if the user explicitly asks for machine-readable output.\n- Use `terms` for the primary search text.\n- Keep `count` modest and page with `offset` instead of pulling large result sets at once.\n\n## Input\n- Read one JSON object from stdin.\n- Required field: `terms`\n- Optional fields: `params`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common NCBI Gene patterns:\n  - `{\"terms\":\"TP53\",\"params\":{\"df\":\"GeneID,Symbol,description\"}}`\n  - `{\"terms\":\"BRCA\",\"params\":{\"count\":10,\"df\":\"chromosome,GeneID,Symbol,description,type_of_gene\"},\"max_items\":10}`\n  - `{\"terms\":\"kinase\",\"params\":{\"count\":10,\"offset\":10,\"df\":\"GeneID,Symbol,description\"},\"max_items\":10}`\n\n## Output\n- Success returns `ok`, `source`, `terms`, `total`, `codes`, `display_rows`, `extra_fields`, and truncation metadata.\n- Use `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"terms\":\"TP53\",\"params\":{\"count\":10,\"df\":\"GeneID,Symbol,description\"},\"max_items\":10}' | python scripts/ncbi_gene_clinicaltables.py\n```\n\n## References\n- No additional runtime references are required; keep the import package limited to this file and `scripts/ncbi_gene_clinicaltables.py`.\n\nFile v1.0.2:skills/ncbi-entrez-skill/SKILL.md\n\n---\nname: ncbi-entrez-skill\ndescription: Submit compact NCBI Entrez E-Utilities requests for PubMed, Gene, Protein, Nucleotide, PMC metadata, and GEO metadata workflows. Use when a user wants concise Entrez search, fetch, summary, or link results; save raw JSON or XML only on request.\n---\n\n## Operating rules\n- Use `scripts/ncbi_entrez.py` for all Entrez calls in this package.\n- Use explicit `endpoint` values such as `esearch`, `esummary`, `efetch`, `elink`, or `einfo`.\n- Search-style Entrez calls are better with `retmax=10` and `max_items=10`.\n- GEO is nested under this skill. Use `db=gds` or `db=geoprofiles` for GEO metadata and load `references/geo.md` only when the user is specifically asking about GEO.\n- BLAST workflows belong in `ncbi-blast-skill`. PMC Open Access workflows belong in `ncbi-pmc-skill`. Datasets v2 workflows belong in `ncbi-datasets-skill`.\n- Re-run requests in long conversations instead of relying on older tool output.\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\n\n## Execution behavior\n- Return concise markdown summaries from the script output by default.\n- Return raw JSON or XML only if the user explicitly asks for machine-readable output.\n- Prefer targeted endpoint calls instead of broad unfiltered dumps.\n- If the user needs the full raw response, set `save_raw=true` and report the saved file path.\n\n## Input\n- Read one JSON object from stdin.\n- Required field: `endpoint`\n- Optional fields: `params`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\n- Common Entrez patterns:\n  - `{\"endpoint\":\"esearch\",\"params\":{\"db\":\"pubmed\",\"term\":\"KRAS AND colorectal cancer\",\"retmode\":\"json\",\"retmax\":10},\"max_items\":10}`\n  - `{\"endpoint\":\"esummary\",\"params\":{\"db\":\"gene\",\"id\":\"7157\",\"retmode\":\"json\"},\"max_items\":10}`\n  - `{\"endpoint\":\"efetch\",\"params\":{\"db\":\"protein\",\"id\":\"NP_000537.3\",\"retmode\":\"xml\"},\"response_format\":\"xml\",\"max_items\":10}`\n  - `{\"endpoint\":\"elink\",\"params\":{\"dbfrom\":\"gds\",\"db\":\"pubmed\",\"id\":\"200000001\",\"retmode\":\"json\"},\"max_items\":10}`\n\n## Output\n- Success returns `ok`, `source`, endpoint metadata, and either compact `records`, a compact `summary`, or `text_head`.\n- Use `raw_output_path` when `save_raw=true`.\n- Failure returns `ok=false` with `error.code` and `error.message`.\n\n## Execution\n```bash\necho '{\"endpoint\":\"esearch\",\"params\":{\"db\":\"gene\",\"term\":\"TP53[gene] AND human[orgn]\",\"retmode\":\"json\",\"retmax\":10},\"max_items\":10}' | python scripts/ncbi_entrez.py\n```\n\n## References\n- Load `references/geo.md` only when the user specifically needs GEO query patterns.\n- Keep the import package limited to this file, `references/geo.md`, and `scripts/ncbi_entrez.py`.\n\nArchive v1.0.1: 8 files, 23034 bytes\n\nFiles: _meta.json (138b), references/drug-naming.md (4091b), references/mcp-tools.md (9492b), references/pharma-intelligence-workflow.md (9308b), references/regulatory-timelines.md (5149b), references/sources-by-region.md (8513b), skill-card.md (3273b), SKILL.md (14419b)\n\nFile v1.0.1:SKILL.md\n\n---\nname: pharma-intelligence\ndescription: >\n  In-depth, multi-region pharmaceutical intelligence search and synthesis,\n  plus drug repurposing, target discovery, clinical evidence review, and\n  bioactivity analysis. Use this skill whenever the user asks about drug\n  approvals, clinical trials, regulatory submissions, pipeline assets, patent\n  landscapes, competitive intelligence, scientific evidence, disease targets,\n  genetic associations, or compound bioactivity for any drug, target,\n  indication, or company — especially when coverage of China, US, Europe,\n  Japan, South Korea, or Australia is needed. Trigger even for casual queries\n  like \"what's the approval status of X in China\", \"find trials for Y in\n  Japan\", \"compare pipeline coverage across regions\", \"find drugs for disease\n  Z\", or \"what targets are associated with condition W\". Always consult this\n  skill before answering any pharma or biomedical research question that\n  requires source-grounded data.\n---\n\n# Global Pharma Intelligence & Biomedical Research Skill\n\nSystematic, source-prioritized search and synthesis across regulatory, clinical,\nacademic, and commercial databases — covering all major pharmaceutical markets\nand 14+ biomedical research databases.\n\n## MCP Server — How to Invoke\n\nThere is no dedicated MCP tool in your toolbox. Call the unified endpoint over HTTP via `web_fetch` (POST) or `run_in_terminal` (curl):\n\n```\nhttps://mcp.sciminer.tech/tools/unified/mcp\n```\n\nEvery call is a JSON-RPC POST. Always set `Content-Type: application/json` and `Accept: application/json`.\n\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"ctg_search_studies\",\"arguments\":{\"intervention\":\"pan-RAS\",\"condition\":\"cancer\",\"max_results\":20}},\"id\":1}'\n```\n\nSee [references/mcp-tools.md](./references/mcp-tools.md) for every tool's parameters and return shape.\n\n---\n\n## Core Principle: Tiered Source Priority\n\nEvery region follows a 3-tier hierarchy. Higher tiers override lower-tier claims; always cite the tier.\n\n| Tier | Type | Description |\n|------|------|-------------|\n| **Tier 1** | Regulatory | Official agency submissions, approvals, labels |\n| **Tier 2** | Trial registries | Prospective/registered clinical evidence |\n| **Tier 3** | Academic / IP | Published papers, conferences, patents |\n\nFor the per-region source map (CN / US / EU / JP / KR / AU + global) with URLs and access notes, see [references/sources-by-region.md](./references/sources-by-region.md).\n\n---\n\n## Search Workflow\n\n### Step 1 — Classify the Query (pick ONE intent)\n\n| # | Intent | Trigger phrases |\n|---|--------|-----------------|\n| A | Trial landscape | \"trials of X\", \"clinical studies of\", \"who is testing\", \"phase 2/3 of\" |\n| B | Approval / regulatory status | \"is X approved\", \"approval status\", \"FDA/EMA/NMPA cleared\" |\n| C | Safety / adverse events | \"side effects of\", \"is X safe\", \"adverse events\", \"black box\" |\n| D | Pipeline / competitive intel | \"pipeline\", \"competitive landscape\", \"who else is developing\" |\n| E | Patent / IP / exclusivity | \"when does patent expire\", \"patent landscape\", \"exclusivity\" |\n| F | Target / mechanism / drug discovery | \"drugs targeting X\", \"mechanism of\", \"bioactivity\", \"IC50\" |\n| G | Repurposing / target discovery | \"repurpose for\", \"targets associated with disease\", \"genetic basis\" |\n| H | Literature / evidence review | \"recent papers on\", \"what's known about\", \"systematic review\" |\n\nAlso capture: **regions in scope** (US / EU / JP / CN / KR / AU / global) and **time horizon**.\n\n### Step 2 — Execute the Per-Intent Sequence\n\nRun the workflow for the chosen intent (see [Per-Intent Workflows](#per-intent-workflows)) in order. For sources without MCP coverage (CN NMPA/CDE, EMA EPAR, PMDA, jRCT, CTIS, CRIS, ANZCTR, Orange Book), use `web_fetch` only at the steps that name them.\n\nResolve identifiers as needed:\n- Free-text disease → MONDO/EFO ID via `opentargets_search`\n- Free-text gene → HGNC symbol via `mygene_search_genes`\n- Cross-database ID conversion → `nodenorm_get_normalized_nodes`\n\n### Step 3 — Resolve Conflicts\n\n1. Higher-tier source wins (Tier 1 > Tier 2 > Tier 3).\n2. More recent data wins within the same tier.\n3. Flag unresolved conflicts; do not silently pick one.\n\n### Step 4 — Synthesize and Present\n\nStructure output to match the intent of the question:\n- Trial landscape → table of trials (NCT/registry ID, phase, status, sponsor, N, primary endpoint).\n- Approval status → region × status × date × indications table.\n- Safety → top FAERS reactions plus black-box / warnings.\n- Pipeline → drug × company × phase × mechanism table.\n- Patent → patent number, jurisdiction, expiry.\n\nAlways cite source, tier, and access date.\n\n---\n\n## Per-Intent Workflows\n\n### A. Trial Landscape\n\n*\"What clinical studies / trials exist for [drug | target | indication]?\"*\n\nDefault scope = ALL regions. Only narrow if the user names a single region.\n\n`ctg_search_studies` covers only ClinicalTrials.gov, which is primarily US-registered trials. Run each regional source in parallel.\n\n1. **United States** — `ctg_search_studies` via MCP.\n   - Use `intervention` for a drug, `condition` for a disease, both for combined.\n   - For a target/class (e.g., \"pan-RAS\", \"PD-L1 inhibitor\"): pass the class term as `intervention` plus a relevant `condition`.\n   - Then `ctg_get_study` on top hits for eligibility, endpoints, sponsor, locations.\n2. **China** — `web_fetch`:\n   - `http://www.chinadrugtrials.org.cn` (mandatory CN IND registry)\n   - `https://www.chictr.org.cn` (ChiCTR, WHO primary)\n3. **Europe** — `web_fetch`:\n   - `https://euclinicaltrials.eu` (CTIS — current EU register)\n   - `https://eudract.ema.europa.eu` (EudraCT — legacy historical trials)\n   - `https://www.isrctn.com` (ISRCTN, UK/global)\n4. **Japan** — `web_fetch`:\n   - `https://jrct.niph.go.jp` (jRCT — mandatory JP registry)\n   - `https://www.umin.ac.jp/ctr/` (UMIN-CTR — legacy)\n5. **South Korea** — `web_fetch` `https://cris.nih.go.kr`.\n6. **Australia / New Zealand** — `web_fetch` `https://www.anzctr.org.au`.\n7. **WHO ICTRP catch-all** — `web_fetch` `https://trialsearch.who.int` for any WHO primary registry (covers India CTRI, Iran IRCT, Brazil ReBEC, etc.). Also `europepmc_search` via MCP for ICTRP-linked publications.\n8. **Published results** — `pubmed_search_articles` with NCT ID or drug name to surface completed-trial papers.\n9. **US company-disclosed pipeline** (optional) — `edgar_fulltext_search` for US-listed sponsors.\n\nFor every regional `web_fetch`: query both INN and brand name; for CN also use the Chinese transliteration (see [references/drug-naming.md](./references/drug-naming.md)). Aggregate results in one table with a \"Registry\" column.\n\n### B. Approval / Regulatory Status\n\n*\"Is [drug] approved in [region]?\"*\n\n1. **US** — `openfda_search_drug_labels` + `dailymed_search_drug_labels` (label date anchors approval); `fda_orphan_search_designations` for orphan status.\n2. **Non-US** — `web_fetch` the regional Tier 1 source (NMPA, EMA EPAR, PMDA, MFDS, TGA). For CN, also search Chinese characters.\n3. `chembl_get_drug_indications` — cross-check approved indications and max phase.\n4. Say \"not approved\" only when Tier 1 affirms denial/withdrawal. Otherwise: \"no record found as of [date]\".\n\n### C. Safety / Adverse Events\n\n1. `openfda_search_adverse_events` (drug_name, `seriousness=serious`).\n2. `openfda_get_drug_label` with `section=\"warnings\"` and `section=\"contraindications\"`.\n3. `chembl_get_molecule` for the black-box warning flag.\n4. `pubmed_search_articles` with `keywords: [\"adverse effect\", \"toxicity\"]` for case reports and post-marketing literature.\n\n### D. Pipeline / Competitive Intelligence\n\n*\"Who else is developing for [indication / target]? What's the global competitive landscape?\"*\n\nDefault scope = ALL regions. A competitive landscape without the active-trial picture is incomplete, so run the full multi-region trial sweep from Workflow A and then layer pipeline-specific sources on top.\n\n1. **Active trials — all regions** — run [Workflow A](#a-trial-landscape) steps 1–7 in full, optionally adding `recruitment_status=RECRUITING` (or `ACTIVE_NOT_RECRUITING`) and a `phase` filter to focus on competitors at a specific stage.\n2. **Company disclosures** — `edgar_fulltext_search` for pipeline language in 10-K / 10-Q / 8-K (US-listed sponsors only).\n3. **Patent activity per company** — `web_fetch` `https://patents.google.com` with an `assignee:` filter (or WIPO PATENTSCOPE / Espacenet — see Workflow E).\n4. **Published results** — `pubmed_search_articles` with NCT IDs or drug names to surface completed-trial papers.\n\nAggregate into one table: drug × company × phase × mechanism × registry/region.\n\n### E. Patent / IP / Exclusivity\n\nAll listed patent sources are free and require no API key.\n\n1. **Global patent search** — `web_fetch` one or more of:\n   - `https://patents.google.com` (Google Patents — best full-text search, covers USPTO, EPO, WIPO, JPO, CNIPA, KIPO).\n   - `https://patentscope.wipo.int` (WIPO PATENTSCOPE — authoritative for PCT applications and national filings worldwide).\n   - `https://worldwide.espacenet.com` (EPO Espacenet — strongest European and family-tree coverage).\n2. **US patents (structured)** — `uspto_ppubs_search_patents` via MCP for granted patents and applications.\n3. **Patent family / cross-jurisdiction equivalents** — Espacenet's \"INPADOC patent family\" view, or Google Patents' \"Worldwide applications\" section.\n4. **Orange Book** (patent + exclusivity expiry for FDA-approved drugs) — `web_fetch` `https://www.accessdata.fda.gov/scripts/cder/ob`.\n5. **Orphan exclusivity** — `fda_orphan_search_exclusivity` (7-year US orphan exclusivity).\n\n### F. Target / Mechanism / Drug Discovery\n\n1. `chembl_find_drugs_by_target` (`target_name` = gene symbol, `include_all_mechanisms=true`).\n2. `chembl_get_mechanism` for each candidate.\n3. `chembl_get_activities` — IC50 / Kd / EC50 for bioactivity comparisons.\n4. `uniprot_search_proteins` — protein function and druggability.\n5. `reactome_search_pathways` or `kegg_find_pathways` — pathway context.\n\n### G. Repurposing / Target Discovery\n\n1. `opentargets_search` (`entity_type=\"disease\"`) → MONDO ID.\n2. `opentargets_get_associations` (`disease_id`, size 20–30) → ranked targets by evidence score.\n3. `gwas_search_associations` — variants linking targets to disease.\n4. `omim_search_entries` — Mendelian basis (requires API key).\n5. For each top target: `chembl_find_drugs_by_target` (`include_all_mechanisms=true`).\n6. `ctg_search_studies` with each drug as intervention for prior-art trials.\n7. `openfda_search_adverse_events` as a safety filter for non-trivial candidates.\n\n### H. Literature / Evidence Review\n\n1. `pubmed_search_articles` — entry point; use `diseases`, `chemicals`, `genes` for entity-aware filtering.\n2. `europepmc_search` — broader: grants, preprints, non-MEDLINE.\n3. `europepmc_search_preprints` — bioRxiv / medRxiv only.\n4. `pubmed_get_article` — abstract or full text for top hits.\n\n---\n\n## Combination Strategies (cross-intent)\n\nUse only when a question genuinely spans multiple intents.\n\n- **Disease → Targets → Drugs → Trials**: `opentargets_search` → `opentargets_get_associations` → `chembl_find_drugs_by_target` → `ctg_search_studies`\n- **Gene → Protein → Pathways → Drugs**: `mygene_search_genes` → `uniprot_get_protein` → `reactome_search_pathways` → `chembl_find_drugs_by_target`\n- **Variant → Gene → Disease → Treatments**: `myvariant_get_variant` → `mygene_get_gene` → `omim_search_entries` → `chembl_find_drugs_by_target`\n- **Drug → Safety → Label → Trials**: `chembl_get_mechanism` → `openfda_search_adverse_events` → `openfda_get_drug_label` → `ctg_search_studies`\n\n---\n\n## API Keys\n\nMost APIs require no key. Exceptions:\n\n| Database | Key | Source |\n|----------|-----|--------|\n| OMIM | Required | https://omim.org/api |\n| NCI Clinical Trials | Optional | https://clinicaltrialsapi.cancer.gov |\n| OpenFDA | Optional (higher rate limits) | https://open.fda.gov/apis |\n\nAll others (ChEMBL, OpenTargets, PubMed, ClinicalTrials.gov, Reactome, KEGG, UniProt, GWAS, Pathway Commons, MyGene / MyVariant / MyChem, Node Normalization, USPTO PPUBS) are public. Patent landscape work uses Google Patents, WIPO PATENTSCOPE, and Espacenet via `web_fetch` — no keys required.\n\n---\n\n## Output Quality Standards\n\n- Never fabricate approval dates, trial IDs, or efficacy numbers.\n- Attribute every claim to its source and tier.\n- Flag gaps explicitly (e.g., \"No registered trials found in jRCT as of [date]\").\n- Distinguish \"no data found\" from \"not approved\" — absence of evidence ≠ negative regulatory decision.\n- For Chinese sources: note whether the search was conducted in Chinese characters; romanization alone may miss records.\n\n---\n\n## Troubleshooting\n\n**No results?**\n- Try alternative terms (INN vs brand name, gene symbol vs protein name).\n- Use standardized IDs: MONDO for diseases, HGNC for genes, ChEMBL IDs for compounds, Ensembl for OpenTargets.\n- Convert IDs across databases with `nodenorm_get_normalized_nodes`.\n\n**Too many results?**\n- Add filters: `max_results`, `phase`, `recruitment_status`, `reviewed` (UniProt).\n- Apply date ranges where supported.\n\n**API key errors?**\n- OMIM requires a key; NCI and OpenFDA accept optional keys for higher rate limits.\n\n**Source not covered by MCP?**\n- Fall back to `web_fetch` for CDE/NMPA, EMA/EPAR, PMDA, jRCT, CTIS, CRIS, ANZCTR, Orange Book.\n\n---\n\n## References\n\n- [references/mcp-tools.md](./references/mcp-tools.md) — Parameters and call format for every MCP tool.\n- [references/drug-naming.md](./references/drug-naming.md) — INN / brand / Chinese / Japanese naming conventions and transliteration.\n- [references/regulatory-timelines.md](./references/regulatory-timelines.md) — Review-clock lengths and milestones per agency (FDA, EMA, PMDA, CDE/NMPA, etc.).\n- [references/sources-by-region.md](./references/sources-by-region.md) — Direct URLs and access notes for all regional regulatory databases.\n- [references/pharma-intelligence-workflow.md](./references/pharma-intelligence-workflow.md) — End-to-end worked example with curl commands (osimertinib in NSCLC).\n\nFile v1.0.1:_meta.json\n\n{\n  \"ownerId\": \"kn725br751g8y5tkj1h6d2krf58356et\",\n  \"slug\": \"pharma-intelligence\",\n  \"version\": \"1.0.1\",\n  \"publishedAt\": 1778489711745\n}\n\nFile v1.0.1:references/drug-naming.md\n\n# Drug Naming Conventions by Region\n\n## Naming Hierarchy (search in this order)\n\n1. **INN** — International Nonproprietary Name (WHO-assigned; universal)\n2. **rINN** — recommended INN (same thing, just confirms WHO adoption)\n3. **USAN** — US Adopted Name (almost always = INN for modern drugs)\n4. **JAN** — Japanese Accepted Name (may differ slightly)\n5. **Brand name** — varies by market and licensee\n6. **CAS number** — Chemical Abstracts Service; best for patent searches\n7. **WHO ATC code** — anatomical/therapeutic/chemical hierarchy\n\n---\n\n## Chinese Drug Name Transliteration Rules\n\nForeign drugs approved in China receive an official Chinese transliteration\nassigned by NMPA/CDE. These are used in all official CN documents.\n\n**Key transliteration patterns:**\n- Small molecule drugs: phonetic transliteration (音译) + type suffix\n  - 替尼 (-tinib, kinase inhibitors), e.g., 伊马替尼 (imatinib)\n  - 西布 (-cib, CDK inhibitors), e.g., 帕博西利 (palbociclib)\n  - 珠单抗 (-zumab), 利单抗 (-limab), 单抗 (monoclonal antibody)\n  - 替雷利珠单抗 (tislelizumab)\n\n**Common antibody suffixes in Chinese:**\n| Suffix class | CN convention | Example |\n|---|---|---|\n| -mab (monoclonal Ab) | 单抗 | nivolumab → 纳武利尤单抗 |\n| -zumab (humanized) | 珠单抗 | bevacizumab → 贝伐珠单抗 |\n| -umab (human) | 尤单抗 | adalimumab → 阿达木单抗 |\n| -limab (PD-1 class) | 利单抗 | sintilimab → 信迪利单抗 |\n\n**Search tip:** Use INN + \"单抗\" or INN + \"替尼\" in CNKI/Wanfang/CDE for\nbetter recall. Also search the full Chinese INN in NMPA databases.\n\n---\n\n## Japanese Drug Name Rules (JAN)\n\nJapanese names follow INN closely but use katakana (カタカナ) transcription:\n- imatinib → イマチニブ\n- trastuzumab → トラスツズマブ\n- pembrolizumab → ペムブロリズマブ\n\n**Search tip:** In J-PlatPat and jRCT, always search katakana form.\nPMDA approval database accepts both INN and brand name in Japanese.\n\n---\n\n## Brand Name Variations by Market (Examples)\n\n| INN | US Brand | EU Brand | JP Brand | CN Brand |\n|-----|----------|----------|----------|----------|\n| imatinib | Gleevec | Glivec | グリベック | 格列卫 |\n| rituximab | Rituxan | MabThera | リツキサン | 美罗华 |\n| pembrolizumab | Keytruda | Keytruda | キイトルーダ | 可瑞达 |\n| nivolumab | Opdivo | Opdivo | オプジーボ | 欧狄沃 |\n| osimertinib | Tagrisso | Tagrisso | タグリッソ | 泰瑞沙 |\n| palbociclib | Ibrance | Ibrance | イブランス | 爱博新 |\n| trastuzumab | Herceptin | Herceptin | ハーセプチン | 赫赛汀 |\n\n> Brand names for biosimilars diverge significantly — always confirm\n> the specific biosimilar product when researching CN/JP markets.\n\n---\n\n## China-Specific Drug Classification Numbers\n\nNMPA approval numbers follow this format: 国药准字 + [category letter] + 8 digits\n\n| Letter | Category |\n|--------|----------|\n| H | 化学药品 — Chemical drugs |\n| Z | 中药 — Traditional Chinese medicine |\n| B | 生物制品 — Biological products |\n| S | 兽药 — Veterinary drugs |\n| J | 进口药品 — Imported drugs (registered in CN) |\n\nExample: 国药准字H20050001 = Chemical drug, approved ~2005\n\n---\n\n## ATC Code System (WHO)\n\nUse ATC codes for systematic indication-based searching:\n\n```\nLevel 1 (Anatomical main group):    L = Antineoplastic\nLevel 2 (Therapeutic subgroup):     L01 = Antineoplastics\nLevel 3 (Pharmacological subgroup): L01X = Other antineoplastics\nLevel 4 (Chemical subgroup):        L01XE = Protein kinase inhibitors\nLevel 5 (Chemical substance):       L01XE01 = Imatinib\n```\n\nATC lookup: https://www.whocc.no/atc_ddd_index/\n\n---\n\n## CAS Number Resources\n\nFor chemical identity across patents and literature:\n- PubChem: https://pubchem.ncbi.nlm.nih.gov (free, comprehensive)\n- ChemSpider: https://www.chemspider.com (cross-reference)\n- SciFinder: (subscription) most authoritative for patents\n\n**When to use CAS:** Patent landscape searches, when INN not yet assigned\n(investigational drugs), structure-activity relationship research.\n\nFile v1.0.1:references/mcp-tools.md\n\n# MCP Tools Quick Reference\n\n## How to Invoke\n\n**There is no dedicated \"MCP tool\" in your toolbox.** Call the endpoint directly using `web_fetch` (HTTP POST) or `run_in_terminal` (curl).\n\n```bash\n# curl — run_in_terminal\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"TOOL_NAME\",\"arguments\":{...}},\"id\":1}'\n```\n\n```\n# web_fetch — HTTP POST\nPOST https://mcp.sciminer.tech/tools/unified/mcp\nContent-Type: application/json\nAccept: application/json\n\n{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"TOOL_NAME\",\"arguments\":{...}},\"id\":1}\n```\n\nOr use individual servers: `https://mcp.sciminer.tech/tools/{server}/mcp`\n\n---\n\n## Drug / Compound Lookup\n\n### `pubchem_search_compound` — Find compound by name/CAS/SMILES\n```json\n{\"name\": \"pubchem_search_compound\", \"arguments\": {\"query\": \"imatinib\", \"max_results\": 5}}\n```\nReturns: CID, IUPAC name, synonyms (brand names), molecular formula, InChIKey\n\n### `mychem_search_drugs` — Cross-database drug lookup\n```json\n{\"name\": \"mychem_search_drugs\", \"arguments\": {\"query\": \"gleevec\", \"max_results\": 5}}\n```\nReturns: ChEMBL ID, DrugBank ID, PubChem CID, SMILES, pharmacology summary in one call\n\n### `chembl_search_molecules` — Search ChEMBL by name/synonym\n```json\n{\"name\": \"chembl_search_molecules\", \"arguments\": {\"query\": \"pembrolizumab\", \"max_results\": 10}}\n```\n\n### `chembl_get_molecule` — Full molecule data by ChEMBL ID\n```json\n{\"name\": \"chembl_get_molecule\", \"arguments\": {\"molecule_id\": \"CHEMBL2397693\"}}\n```\nReturns: SMILES, MW, LogP, ATC codes, max phase, indications, black box warnings\n\n### `chembl_find_drugs_by_indication` — All drugs approved/investigated for a disease\n```json\n{\"name\": \"chembl_find_drugs_by_indication\", \"arguments\": {\"indication\": \"non-small cell lung cancer\", \"max_results\": 30}}\n```\n\n### `chembl_find_drugs_by_target` — All drugs targeting a gene/protein\n```json\n{\"name\": \"chembl_find_drugs_by_target\", \"arguments\": {\"target_name\": \"EGFR\", \"include_all_mechanisms\": true, \"max_results\": 20}}\n```\n\n### `chembl_get_mechanism` / `chembl_get_molecule_mechanisms` — Mechanism of action\n```json\n{\"name\": \"chembl_get_mechanism\", \"arguments\": {\"molecule_id\": \"CHEMBL2397693\"}}\n```\n\n### `chembl_get_drug_indications` — All approved + investigational indications for a drug\n```json\n{\"name\": \"chembl_get_drug_indications\", \"arguments\": {\"molecule_id\": \"CHEMBL2397693\"}}\n```\n\n---\n\n## US Regulatory\n\n### `openfda_search_drug_labels` — FDA drug labels (package inserts)\n```json\n{\"name\": \"openfda_search_drug_labels\", \"arguments\": {\"drug_name\": \"pembrolizumab\", \"section\": \"warnings\"}}\n```\n\n### `openfda_get_drug_label` — Full label by set_id\n```json\n{\"name\": \"openfda_get_drug_label\", \"arguments\": {\"set_id\": \"abc123\", \"section\": \"indications_and_usage\"}}\n```\nSections: `warnings`, `contraindications`, `adverse_reactions`, `dosage_and_administration`, `indications_and_usage`\n\n### `openfda_search_adverse_events` — FAERS adverse event search\n```json\n{\"name\": \"openfda_search_adverse_events\", \"arguments\": {\"drug_name\": \"nivolumab\", \"seriousness\": \"serious\", \"limit\": 50}}\n```\n\n### `dailymed_search_drug_labels` — Official FDA label database (SPL)\n```json\n{\"name\": \"dailymed_search_drug_labels\", \"arguments\": {\"drug_name\": \"keytruda\", \"max_results\": 5}}\n```\n\n### `dailymed_get_spl` — Full structured product label\n```json\n{\"name\": \"dailymed_get_spl\", \"arguments\": {\"set_id\": \"spl-set-id\"}}\n```\n\n### `fda_orphan_search_designations` — Orphan drug designations (OOPD)\n```json\n{\"name\": \"fda_orphan_search_designations\", \"arguments\": {\"drug_name\": \"imatinib\", \"max_results\": 10}}\n```\n\n### `fda_orphan_search_exclusivity` — 7-year orphan exclusivity periods\n```json\n{\"name\": \"fda_orphan_search_exclusivity\", \"arguments\": {\"drug_name\": \"imatinib\"}}\n```\n\n---\n\n## Clinical Trials\n\n### `ctg_search_studies` — ClinicalTrials.gov search with filters\n```json\n{\n  \"name\": \"ctg_search_studies\",\n  \"arguments\": {\n    \"condition\": \"non-small cell lung cancer\",\n    \"intervention\": \"osimertinib\",\n    \"recruitment_status\": \"RECRUITING\",\n    \"phase\": \"PHASE3\",\n    \"max_results\": 20\n  }\n}\n```\n`recruitment_status` values: `RECRUITING`, `ACTIVE_NOT_RECRUITING`, `COMPLETED`, `TERMINATED`, `WITHDRAWN`\n`phase` values: `PHASE1`, `PHASE2`, `PHASE3`, `PHASE4`, `EARLY_PHASE1`\n\n### `ctg_get_study` — Full trial protocol\n```json\n{\"name\": \"ctg_get_study\", \"arguments\": {\"nct_id\": \"NCT04604678\"}}\n```\nReturns: eligibility criteria, endpoints, locations, enrollment, sponsor\n\n### `nci_search_trials` — NCI oncology-specific trial search (requires API key)\n```json\n{\"name\": \"nci_search_trials\", \"arguments\": {\"condition\": \"glioblastoma\", \"api_key\": \"YOUR_KEY\", \"max_results\": 20}}\n```\n\n---\n\n## Patent / IP\n\n### Global patent search (no MCP — use `web_fetch`)\n\nAll free and require no API key:\n\n- **Google Patents** — `https://patents.google.com/?q=<query>` (best full-text; covers USPTO, EPO, WIPO, JPO, CNIPA, KIPO).\n- **WIPO PATENTSCOPE** — `https://patentscope.wipo.int/search/en/result.jsf?query=<query>` (authoritative for PCT and national filings).\n- **Espacenet (EPO)** — `https://worldwide.espacenet.com/patent/search?q=<query>` (best European coverage; use INPADOC patent-family view for cross-jurisdiction equivalents).\n\nFilter by assignee/applicant via the source's UI parameters (e.g., Google Patents `assignee:` keyword).\n\n### `uspto_ppubs_search_patents` — US-only granted patents\n```json\n{\"name\": \"uspto_ppubs_search_patents\", \"arguments\": {\"query\": \"anti-PD-1 antibody\", \"max_results\": 20}}\n```\n\n### `uspto_ppubs_search_applications` — US pending applications\n```json\n{\"name\": \"uspto_ppubs_search_applications\", \"arguments\": {\"query\": \"EGFR inhibitor cancer\", \"max_results\": 20}}\n```\n\n---\n\n## Financial / Competitive Intelligence\n\n### `edgar_search_company` — Find company in EDGAR\n```json\n{\"name\": \"edgar_search_company\", \"arguments\": {\"query\": \"Merck\", \"max_results\": 10}}\n```\n\n### `edgar_search_filings` — Search SEC filings by type\n```json\n{\"name\": \"edgar_search_filings\", \"arguments\": {\"company_name\": \"Merck\", \"form_type\": \"10-K\", \"max_results\": 5}}\n```\n`form_type`: `10-K` (annual), `10-Q` (quarterly), `8-K` (current events/FDA decisions)\n\n### `edgar_fulltext_search` — Search full text of SEC filings\n```json\n{\"name\": \"edgar_fulltext_search\", \"arguments\": {\"query\": \"pembrolizumab pipeline approval\", \"max_results\": 10}}\n```\nBest for: finding pipeline disclosures, FDA decision announcements in filings\n\n---\n\n## Literature\n\n### `pubmed_search_articles` — PubMed literature search\n```json\n{\n  \"name\": \"pubmed_search_articles\",\n  \"arguments\": {\n    \"query\": \"osimertinib resistance mechanisms\",\n    \"diseases\": [\"non-small cell lung cancer\"],\n    \"chemicals\": [\"osimertinib\"],\n    \"max_results\": 30\n  }\n}\n```\n\n### `europepmc_search` — Broader search (preprints + non-MEDLINE + grants)\n```json\n{\"name\": \"europepmc_search\", \"arguments\": {\"query\": \"PD-L1 immunotherapy 2025\", \"max_results\": 20}}\n```\n\n### `europepmc_search_preprints` — bioRxiv/medRxiv preprints only\n```json\n{\"name\": \"europepmc_search_preprints\", \"arguments\": {\"query\": \"CAR-T cell therapy GvHD\", \"max_results\": 10}}\n```\n\n---\n\n## Target / Biology\n\n### `opentargets_search` — Find disease (MONDO ID) or target (Ensembl ID)\n```json\n{\"name\": \"opentargets_search\", \"arguments\": {\"query\": \"non-small cell lung cancer\", \"entity_type\": \"disease\"}}\n```\n\n### `opentargets_get_associations` — Targets ranked by evidence score (0–1)\n```json\n{\"name\": \"opentargets_get_associations\", \"arguments\": {\"disease_id\": \"MONDO_0005233\", \"size\": 25}}\n```\n\n### `opentargets_get_evidence` — Evidence breakdown for target-disease pair\n```json\n{\"name\": \"opentargets_get_evidence\", \"arguments\": {\"target_id\": \"ENSG00000146648\", \"disease_id\": \"MONDO_0005233\"}}\n```\n\n### `uniprot_search_proteins` — Search proteins by name/gene/function\n```json\n{\"name\": \"uniprot_search_proteins\", \"arguments\": {\"query\": \"EGFR\", \"reviewed\": true, \"max_results\": 5}}\n```\n`reviewed: true` returns only Swiss-Prot curated entries\n\n### `reactome_search_pathways` — Find relevant pathways\n```json\n{\"name\": \"reactome_search_pathways\", \"arguments\": {\"query\": \"EGFR signaling\", \"max_results\": 10}}\n```\n\n### `reactome_get_disease_pathways` — Disease-specific pathways\n```json\n{\"name\": \"reactome_get_disease_pathways\", \"arguments\": {\"disease_name\": \"lung cancer\"}}\n```\n\n### `gwas_search_associations` — Genetic variants linked to disease\n```json\n{\"name\": \"gwas_search_associations\", \"arguments\": {\"query\": \"non-small cell lung cancer\", \"p_upper\": 0.00001}}\n```\n\n### `omim_search_entries` — Genetic disease basis (requires API key)\n```json\n{\"name\": \"omim_search_entries\", \"arguments\": {\"search_term\": \"lung adenocarcinoma\", \"api_key\": \"YOUR_KEY\"}}\n```\n\n### `nodenorm_get_normalized_nodes` — Convert IDs across databases\n```json\n{\"name\": \"nodenorm_get_normalized_nodes\", \"arguments\": {\"curie\": \"HGNC:3236\"}}\n```\nReturns: NCBI Gene, UniProt, Ensembl, MyGene equivalents in one call\n\n### `mygene_search_genes` — Gene info (location, aliases, cross-references)\n```json\n{\"name\": \"mygene_search_genes\", \"arguments\": {\"query\": \"EGFR\", \"max_results\": 5}}\n```\n\n### `myvariant_get_variant` — Variant effects, ClinVar significance\n```json\n{\"name\": \"myvariant_get_variant\", \"arguments\": {\"variant_id\": \"rs121913529\"}}\n```\nReturns: SIFT/PolyPhen scores, ClinVar significance, gnomAD frequency, CADD score\n\nFile v1.0.1:references/pharma-intelligence-workflow.md\n\n# Pharma Intelligence Workflow — Complete Example\n\n**Goal**: Full competitive intelligence report on a drug in a target indication across regions.\n**Example**: Osimertinib (Tagrisso) in EGFR-mutant NSCLC\n\n---\n\n## Step 1: Identify the Drug\n\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"chembl_search_molecules\",\"arguments\":{\"query\":\"osimertinib\",\"max_results\":5}},\"id\":1}'\n```\n\nNote the ChEMBL ID (e.g., `CHEMBL3353410`) and max phase. Then get full details:\n\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"chembl_get_molecule\",\"arguments\":{\"molecule_id\":\"CHEMBL3353410\"}},\"id\":2}'\n```\n\nReturns: ATC codes, approved indications, mechanism, black box warnings.\n\n---\n\n## Step 2: US Regulatory Status\n\n### FDA label (approved indications + label date)\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"openfda_search_drug_labels\",\"arguments\":{\"drug_name\":\"osimertinib\",\"section\":\"indications_and_usage\"}},\"id\":3}'\n```\n\n### Orphan designation (if applicable)\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"fda_orphan_search_designations\",\"arguments\":{\"drug_name\":\"osimertinib\",\"max_results\":5}},\"id\":4}'\n```\n\n### Orange Book / exclusivity → use web_search for:\n> `site:accessdata.fda.gov/scripts/cder/ob osimertinib`\n\n---\n\n## Step 3: US Clinical Trials (Phase 3 + Active)\n\n> **Scope:** `ctg_search_studies` queries ClinicalTrials.gov, which is **US-only**. For CN/EU/JP/KR/AU trials, run the regional `web_fetch` calls in Step 9 in parallel.\n\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"ctg_search_studies\",\"arguments\":{\"condition\":\"non-small cell lung cancer\",\"intervention\":\"osimertinib\",\"phase\":\"PHASE3\",\"max_results\":20}},\"id\":5}'\n```\n\nGet full protocol for key trials:\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"ctg_get_study\",\"arguments\":{\"nct_id\":\"NCT02151981\"}},\"id\":6}'\n```\n\n---\n\n## Step 4: Safety Profile (FAERS)\n\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"openfda_search_adverse_events\",\"arguments\":{\"drug_name\":\"osimertinib\",\"seriousness\":\"serious\",\"limit\":50}},\"id\":7}'\n```\n\n---\n\n## Step 5: Patent / IP Landscape\n\nGlobal patent search uses free web sources (no API key):\n\n- Google Patents: `web_fetch https://patents.google.com/?q=osimertinib+EGFR+T790M`\n- WIPO PATENTSCOPE: `web_fetch https://patentscope.wipo.int/search/en/result.jsf?query=osimertinib`\n- Espacenet (EPO): `web_fetch https://worldwide.espacenet.com/patent/search?q=osimertinib` (use the \"INPADOC patent family\" view for cross-jurisdiction equivalents)\n\nStructured US patents via MCP:\n\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"uspto_ppubs_search_patents\",\"arguments\":{\"query\":\"osimertinib EGFR inhibitor\",\"max_results\":20}},\"id\":8}'\n```\n\nFor Orange Book expiry: `web_fetch https://www.accessdata.fda.gov/scripts/cder/ob`.\n\n---\n\n## Step 6: Competitive Landscape (Same Indication)\n\n### All drugs approved/investigated for NSCLC\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"chembl_find_drugs_by_indication\",\"arguments\":{\"indication\":\"non-small cell lung cancer\",\"max_results\":50}},\"id\":10}'\n```\n\n### All active NSCLC trials (any intervention)\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"ctg_search_studies\",\"arguments\":{\"condition\":\"non-small cell lung cancer\",\"recruitment_status\":\"RECRUITING\",\"phase\":\"PHASE3\",\"max_results\":30}},\"id\":11}'\n```\n\n### Competitor SEC pipeline disclosures\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"edgar_fulltext_search\",\"arguments\":{\"query\":\"EGFR inhibitor NSCLC pipeline 2025\",\"max_results\":10}},\"id\":12}'\n```\n\n---\n\n## Step 7: Target Biology (Mechanism Validation)\n\n### Get EGFR gene info and cross-references\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"mygene_search_genes\",\"arguments\":{\"query\":\"EGFR\",\"max_results\":3}},\"id\":13}'\n```\n\n### Find all drugs targeting EGFR\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"chembl_find_drugs_by_target\",\"arguments\":{\"target_name\":\"EGFR\",\"include_all_mechanisms\":true,\"max_results\":30}},\"id\":14}'\n```\n\n### OpenTargets: NSCLC targets ranked by evidence\n```bash\n# First get disease ID\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"opentargets_search\",\"arguments\":{\"query\":\"non-small cell lung cancer\",\"entity_type\":\"disease\"}},\"id\":15}'\n\n# Then get top targets (use MONDO ID from above)\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"opentargets_get_associations\",\"arguments\":{\"disease_id\":\"MONDO_0005233\",\"size\":20}},\"id\":16}'\n```\n\n---\n\n## Step 8: Literature Review\n\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"pubmed_search_articles\",\"arguments\":{\"query\":\"osimertinib resistance mechanisms 2024 2025\",\"chemicals\":[\"osimertinib\"],\"diseases\":[\"non-small cell lung cancer\"],\"max_results\":30}},\"id\":17}'\n```\n\nFor preprints (ahead of peer review):\n```bash\ncurl -X POST https://mcp.sciminer.tech/tools/unified/mcp \\\n  -H \"Content-Type: application/json\" -H \"Accept: application/json\" \\\n  -d '{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"europepmc_search_preprints\",\"arguments\":{\"query\":\"osimertinib EGFR 2025\",\"max_results\":10}},\"id\":18}'\n```\n\n---\n\n## Step 9: Non-US Regulatory (web_search fallback)\n\nFor regions without MCP coverage, use `web_search` or `web_fetch`:\n\n| Region | Query pattern |\n|--------|---------------|\n| China (NMPA) | `site:nmpa.gov.cn \"奥希替尼\"` or `web_fetch https://www.nmpa.gov.cn/datasearch/...` |\n| Japan (PMDA) | `site:pmda.go.jp osimertinib` or search PMDA approval DB in Japanese (オシメルチニブ) |\n| EU (EMA EPAR) | `web_fetch https://www.ema.europa.eu/en/medicines/human/EPAR/tagrisso` |\n| South Korea (MFDS) | `web_search site:mfds.go.kr osimertinib` |\n| Australia (TGA) | `web_search site:tga.gov.au osimertinib ARTG` |\n\nAlways cross-check approval date against first-approval country to estimate regulatory lag.\n\n---\n\n## Synthesis Template\n\nAfter running the above steps, structure the output as:\n\n```\n## [Drug Name] — [Indication] Intelligence Report\n\n### Regulatory Status\n| Region | Status | Date | Indications |\n|--------|--------|------|-------------|\n| US (FDA) | Approved | YYYY-MM | ... |\n| EU (EMA) | Approved | YYYY-MM | ... |\n| Japan (PMDA) | Approved | YYYY-MM | ... |\n| China (NMPA) | Approved | YYYY-MM | ... |\n\n### Clinical Trials (Active Phase 3)\n| NCT ID | Title | Phase | Status | N | Primary Endpoint |\n|--------|-------|-------|--------|---|-----------------|\n\n### Safety Profile (Top FAERS signals)\n| Reaction | Count | Serious % | Outcome |\n|----------|-------|-----------|---------|\n\n### Patent Landscape\n| Patent | Expiry | Jurisdiction | Coverage |\n|--------|--------|-------------|---------|\n\n### Competitive Landscape (Same Class)\n| Drug | Company | Phase | Mechanism difference |\n|------|---------|-------|---------------------|\n\n### Sources Used (with Tier)\n- Tier 1: FDA label (Drugs@FDA, accessed YYYY-MM-DD)\n- Tier 1: NMPA approval (nmpa.gov.cn, accessed YYYY-MM-DD)\n- Tier 2: ClinicalTrials.gov (ctg_search_studies, accessed YYYY-MM-DD)\n- Tier 3: PubMed (pubmed_search_articles, accessed YYYY-MM-DD)\n```\n\nFile v1.0.1:references/regulatory-timelines.md\n\n# Regulatory Timelines by Agency\n\nReference for setting expectations when estimating approval milestones or\nidentifying gaps between submission and approval in different markets.\n\n---\n\n## Standard Review Timelines\n\n| Agency | Review Type | Clock (months) | Notes |\n|--------|-------------|----------------|-------|\n| **FDA** | Standard NDA/BLA | 12 | From submission (PDUFA date) |\n| **FDA** | Priority Review | 6 | Priority designation required |\n| **FDA** | Breakthrough Therapy | 6 (priority clock) | + intensive guidance |\n| **FDA** | Accelerated Approval | Variable | Based on surrogate endpoint |\n| **EMA** | Centralized Standard | 210 days (+clock stops) | Typically 12–15 months total |\n| **EMA** | PRIME (accelerated) | 150 days active | Early interaction program |\n| **EMA** | Conditional MA | Faster | Post-marketing obligations |\n| **PMDA** | Standard | 12 | From filing date |\n| **PMDA** | Priority Review | 9 | Serious/unmet need |\n| **CDE/NMPA** | Standard NDA | 12–18 | Clock starts at acceptance |\n| **CDE/NMPA** | Priority Review | 6 | National urgent need |\n| **CDE/NMPA** | Conditional Approval | Variable | Post-marketing required |\n| **MFDS** | Standard | 12 | From submission acceptance |\n| **MFDS** | Fast Track | 6 | Serious/life-threatening |\n| **TGA** | Standard | 12 months | From submission |\n| **TGA** | Priority | 6 months | Life-threatening conditions |\n| **MHRA** | Standard | 150 days | Post-Brexit, UK only |\n| **MHRA** | Innovative Licensing | Variable | ILAP pathway |\n\n---\n\n## Key Regulatory Milestones (US / FDA)\n\n```\nPre-IND meeting\nIND submission → 30-day FDA review (clinical hold or proceed)\nPhase 1 → Phase 2 → Phase 3 (typical; may compress)\nEnd of Phase 2 meeting (Type B)\nPre-NDA/BLA meeting (Type B)\nNDA/BLA submission (eCTD format)\n  → Day 60: Filing/Refuse to File decision\n  → Advisory Committee meeting (if convened)\n  → Day 180: Mid-cycle review\n  → PDUFA date: Action (Approval / CRL / Complete Response Letter)\nPost-approval: REMS (if required), label negotiations, manufacturing inspections\n```\n\n## Key Regulatory Milestones (China / CDE-NMPA)\n\n```\nIND application → CDE review (60 working days default)\nClinical trial authorization (CTA)\n  → Phase I (first-in-human): often at dedicated CROs in CN\n  → Phase II/III: registration trials\nPre-NDA communication meeting (沟通交流会)\nNDA submission (申请上市许可)\n  → Acceptance and assignment\n  → Technical review (技术审评)\n  → On-site inspection (核查)\n  → NMPA approval (批准上市)\nPriority Review (优先审评审批): 6-month target\nConditional Approval (附条件批准): post-marketing studies required\nNRDL listing negotiation (国家医保目录谈判): pricing/reimbursement after approval\n```\n\n## Key Regulatory Milestones (Japan / PMDA)\n\n```\nPre-consultation (事前相談)\nClinical trial notification (治験届) — mandatory 30 days before Phase 1\nPhase I → II → III\nNew Drug Application (新薬承認申請 / J-NDA)\n  → PMDA review + inspection\n  → MHLW approval (承認)\n  → NHI price listing (薬価収載) — typically 60–90 days post-approval\n  → Market launch\n```\n\n---\n\n## Expedited Pathways Comparison\n\n| Feature | FDA | EMA | PMDA | CDE/NMPA | MFDS |\n|---------|-----|-----|------|----------|------|\n| Fast Track | ✓ | — | ✓ (優先審査) | ✓ (优先审评) | ✓ |\n| Breakthrough Therapy | ✓ | — | — | ✓ (突破性治疗) | — |\n| Accelerated Approval | ✓ | Conditional MA | ✓ | Conditional (附条件) | ✓ |\n| Priority Medicines | — | PRIME | — | — | — |\n| Orphan Drug | ✓ | ✓ | ✓ | ✓ (2020+) | ✓ |\n| Pediatric Priority | ✓ | PUMA | ✓ | ✓ | — |\n| Rolling Review | ✓ | ✓ (COVID era) | ✓ | ✓ | ✓ |\n\n---\n\n## Typical Lag: US Approval → Ex-US Approval\n\nBased on historical patterns for innovative drugs (oncology/rare disease):\n\n| Market | Median Lag vs FDA | Notes |\n|--------|-------------------|-------|\n| EU (EMA) | 6–18 months | May co-develop; sometimes simultaneous |\n| Japan (PMDA) | 12–24 months | \"Drug lag\" historically; improving |\n| China (NMPA) | 12–36 months (shrinking) | Domestic innovation approvals now often faster |\n| Korea (MFDS) | 12–24 months | Often follows US/EU data |\n| Australia (TGA) | 6–18 months | Sometimes faster than EU |\n\n> Since ~2018, China's CDE reform has dramatically shortened this lag for priority drugs.\n> Monitor domestic CN-first approvals (e.g., EGFR, ALK inhibitors) where China may approve ahead of or concurrently with FDA.\n\n---\n\n## Data Exclusivity Periods by Region\n\n| Region | NCE Exclusivity | Biologic Exclusivity | Orphan |\n|--------|----------------|----------------------|--------|\n| US | 5 years | 12 years | 7 years |\n| EU | 8+2(+1) years | 10 years | 10 years |\n| Japan | 8 years | 10 years | 10 years |\n| China | Not yet harmonized | Limited framework | 7 years (pilot) |\n| Korea | 6 years | — | 10 years |\n| Australia | 5 years | 5 years | 5 years |\n\n> \"8+2+1\" EU formula: 8 yr data exclusivity + 2 yr market exclusivity + 1 yr new indication.\n> CN data exclusivity rules are evolving post-TRIPS Agreement implementation.\n\nFile v1.0.1:references/sources-by-region.md\n\n# Sources by Region — URLs & Access Notes\n\n## 🇨🇳 China\n\n| Source | URL | Notes |\n|--------|-----|-------|\n| CDE | https://www.cde.org.cn | Drug evaluation center. Search by product name or approval number. Chinese-language primary. |\n| NMPA | https://www.nmpa.gov.cn | Final approval body. Drug approval database at https://www.nmpa.gov.cn/datasearch/home-index.html |\n| chinadrugtrials | http://www.chinadrugtrials.org.cn | Mandatory IND registry. Search by drug name (CN or EN), sponsor, or CTR number. |\n| ChiCTR | https://www.chictr.org.cn | WHO primary registry. Better English search interface. |\n| ClinicalTrials.gov (CN) | https://clinicaltrials.gov | Filter: Country = China |\n| CNKI | https://www.cnki.net | Chinese academic literature. Subscription required for full text. |\n| Wanfang | https://www.wanfangdata.com.cn | Alternative to CNKI; strong biomedical coverage. |\n| CNIPA | https://www.cnipa.gov.cn | Chinese patent authority. Search tool: https://pss-system.cponline.cnipa.gov.cn |\n| SinoMed | https://www.sinomed.ac.cn | Chinese biomedical literature, open access. |\n\n**Tips for China searches:**\n- Always search both simplified Chinese and English terms\n- CDE drug numbers follow format: 国药准字 + letter + 8 digits (e.g., 国药准字H20050001)\n- CTR numbers (chinadrugtrials): CTR + YYYYNNNNNN format\n- NMPA approval categories: 化药 (chemical), 生物制品 (biological), 中药 (TCM)\n\n---\n\n## 🇺🇸 United States\n\n| Source | URL | Notes |\n|--------|-----|-------|\n| Drugs@FDA | https://www.accessdata.fda.gov/scripts/cder/daf/ | Full NDA/BLA/ANDA history, labels, reviews |\n| FDA CDER | https://www.fda.gov/drugs | IND policy, drug shortage, safety communications |\n| FDA CBER | https://www.fda.gov/vaccines-blood-biologics | Cell/gene therapy, vaccines, blood products |\n| Orange Book | https://www.accessdata.fda.gov/scripts/cder/ob/ | Patent & exclusivity listings for approved drugs |\n| Purple Book | https://purplebooksearch.fda.gov | Reference biologics + biosimilar interchangeability |\n| SEC EDGAR | https://www.sec.gov/cgi-bin/browse-edgar | Pipeline disclosures in 10-K, 8-K, 20-F filings |\n| ClinicalTrials.gov | https://clinicaltrials.gov | Mandatory US registration. Advanced search supports NCT filter |\n| PubMed | https://pubmed.ncbi.nlm.nih.gov | MEDLINE + PubMed Central. Use MeSH terms for precision |\n| USPTO | https://www.uspto.gov | US patent search; use PatFT/AppFT |\n| Google Patents | https://patents.google.com | Easier cross-jurisdiction search |\n\n**Tips for US searches:**\n- NDA/BLA approval letters downloadable from Drugs@FDA\n- Orange Book: check both patent expiry AND exclusivity dates (can differ by years)\n- EDGAR full-text search: https://efts.sec.gov/LATEST/search-index?q=\"drug name\"\n\n---\n\n## 🇪🇺 Europe\n\n| Source | URL | Notes |\n|--------|-----|-------|\n| EMA EPAR | https://www.ema.europa.eu/en/medicines/download-medicine-data | European Public Assessment Reports with full review history |\n| EMA Product DB | https://www.ema.europa.eu/en/medicines | Status: authorised / refused / withdrawn |\n| CTIS | https://euclinicaltrials.eu | New EU Clinical Trials Information System (replacing EudraCT 2025) |\n| EudraCT (legacy) | https://eudract.ema.europa.eu | Still accessible for historical trials pre-CTIS |\n| MHRA (UK) | https://products.mhra.gov.uk | UK-specific approvals post-Brexit |\n| ISRCTN | https://www.isrctn.com | UK/global registry, WHO-recognized |\n| BfArM (Germany) | https://www.bfarm.de | National agency; also handles parallel import |\n| ANSM (France) | https://ansm.sante.fr | French national agency |\n| Swissmedic | https://www.swissmedic.ch | Switzerland — not EMA member |\n| EPO / Espacenet | https://worldwide.espacenet.com | European patent search, 100+ countries |\n| Embase | https://www.embase.com | Subscription. Superior drug adverse event indexing vs PubMed |\n\n**Tips for EU searches:**\n- EPAR includes refused applications and withdrawn products — check those too\n- CTIS/EudraCT numbers: 2021-NNNNNN-NN format\n- Post-Brexit: MHRA approvals are separate from EMA; UK often approves independently now\n\n---\n\n## 🇯🇵 Japan\n\n| Source | URL | Notes |\n|--------|-----|-------|\n| PMDA Approval DB | https://www.pmda.go.jp/PmdaSearch/iyakuSearch/ | Japanese drug approval search |\n| PMDA Review Reports | https://www.pmda.go.jp/review-services/drug-reviews/review-information/p-drugs/0028.html | Public assessment reports (some English) |\n| jRCT | https://jrct.niph.go.jp | Japan Registry of Clinical Trials (mandatory since 2018) |\n| UMIN-CTR | https://www.umin.ac.jp/ctr/ | University Medical Information Network — legacy, still active |\n| J-STAGE | https://www.jstage.jst.go.jp | Japanese scientific literature, open access |\n| J-PlatPat | https://j-platpat.inpit.go.jp | Japanese patent full-text search |\n| MHLW | https://www.mhlw.go.jp | Ministry of Health guidelines and policy |\n\n**Tips for Japan searches:**\n- PMDA uses Japanese drug names (カタカナ for foreign drugs); search both\n- jRCT IDs: jRCT + number (e.g., jRCT2031190001)\n- PMDA review reports increasingly available in English for key products\n- Japan approval (承認) ≠ listed on NHI price list (収載) — check both for market access\n\n---\n\n## 🇰🇷 South Korea\n\n| Source | URL | Notes |\n|--------|-----|-------|\n| MFDS | https://www.mfds.go.kr | Ministry of Food and Drug Safety |\n| MFDS PharmNet | https://nedrug.mfds.go.kr | Drug product approval database (Korean) |\n| DUR system | https://www.health.kr | Drug utilization review, interaction database |\n| CRIS | https://cris.nih.go.kr | Clinical Research Information Service (WHO-recognized) |\n| RISS | https://www.riss.kr | Korean academic literature |\n| KIPO | https://www.kipo.go.kr | Korean Intellectual Property Office (patents) |\n\n**Tips for Korea searches:**\n- MFDS uses Korean drug names; use English INN + \"허가\" (approval) for searches\n- CRIS IDs: KCT + number\n- Korea has a fast-track designation (신속심사) similar to FDA Breakthrough Therapy\n\n---\n\n## 🇦🇺 Australia / 🇳🇿 New Zealand\n\n| Source | URL | Notes |\n|--------|-----|-------|\n| TGA | https://www.tga.gov.au | Therapeutic Goods Administration |\n| ARTG | https://www.tga.gov.au/resources/artg | Australian Register of Therapeutic Goods (public search) |\n| TGA Public Summary | https://www.tga.gov.au/resources/auspar | Australian Public Assessment Reports (AusPAR) |\n| ANZCTR | https://www.anzctr.org.au | Australia New Zealand Clinical Trials Registry (WHO-recognized) |\n| Medsafe (NZ) | https://www.medsafe.govt.nz | New Zealand regulatory database |\n| IP Australia | https://www.ipaustralia.gov.au | Australian patent database |\n\n**Tips for AU/NZ searches:**\n- TGA uses ARTG number as unique identifier\n- ARTG categories: Prescription (Rx), OTC, Biologicals\n- AusPAR = AU equivalent of EPAR; very detailed, publicly available\n- ANZCTR IDs: ACTRN + 14 digits\n\n---\n\n## 🌐 Cross-Regional & Global\n\n| Source | URL | Notes |\n|--------|-----|-------|\n| WHO ICTRP | https://trialsearch.who.int | Aggregates all WHO primary registries worldwide |\n| WHO Prequalification | https://extranet.who.int/pqweb | Developing market approval decisions |\n| ICH | https://www.ich.org | Harmonized guidelines (E6, E9, M4, etc.) |\n| Cochrane Library | https://www.cochranelibrary.com | Systematic reviews and meta-analyses |\n| Lens.org | https://www.lens.org | Open patent + literature cross-search |\n| Cortellis / Pharmaprojects | (subscription) | Commercial pipeline database — most comprehensive |\n| GlobalData | (subscription) | Commercial competitive intelligence |\n| Citeline (Citeline/Informa) | (subscription) | Trial and pipeline data |\n| AdisInsight | https://adisinsight.springer.com | Partially open; drug pipeline, clinical trial summaries |\n| OpenFDA | https://open.fda.gov | Machine-readable FDA data via API |\n| ClinicalTrials.gov API | https://clinicaltrials.gov/api/v2 | Structured query access to CT.gov |\n\n---\n\n## Drug Naming Cross-Reference Strategy\n\nWhen searching across regions, always check:\n1. **INN** (International Nonproprietary Name) — primary search key\n2. **Brand names** — vary by market (e.g., same molecule, different brands in CN vs US)\n3. **CAS Registry Number** — chemical identity, useful for patents\n4. **WHO ATC code** — anatomical-therapeutic-chemical classification\n5. **Chinese transliteration** — e.g., nivolumab → 纳武利尤单抗\n\nKey INN resource: WHO INN list — https://www.who.int/teams/health-product-and-policy-standards/inn\n\nFile v1.0.1:skill-card.md\n\n## Description: <br>\nProvides source-prioritized pharmaceutical intelligence search and synthesis across regulatory, clinical, academic, patent, and biomedical research sources for drug approvals, trials, pipelines, safety, targets, repurposing, and bioactivity questions. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[sciminer](https://clawhub.ai/user/sciminer) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers, analysts, and research teams use this skill to build source-grounded pharma and biomedical intelligence answers across regions, including approval status, clinical trial landscapes, competitive pipelines, patent/exclusivity checks, safety review, target discovery, and evidence synthesis. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: Biomedical and pharmaceutical research queries, and possible API keys, may be sent to the third-party sciminer.tech endpoint. <br>\nMitigation: Do not include patient data, unpublished research, confidential commercial strategy, proprietary compound details, or real API keys unless the endpoint has been independently trusted and its handling of request contents and credentials is understood. <br>\nRisk: The security scan verdict is suspicious because privacy and credential-handling boundaries are not clear enough for broad biomedical and pharma queries. <br>\nMitigation: Install only when users are comfortable with that third-party routing, and keep sensitive credentials and confidential research content out of requests by default. <br>\nRisk: Regulatory, clinical, patent, and biomedical results can be incomplete or conflicting across regions and source tiers. <br>\nMitigation: Prefer official regulatory and registry sources, cite the source tier and access date, and explicitly flag unresolved conflicts or gaps instead of treating no results as a definitive negative finding. <br>\n\n\n## Reference(s): <br>\n- [ClawHub Skill Page](https://clawhub.ai/sciminer/pharma-intelligence) <br>\n- [MCP Tools Quick Reference](references/mcp-tools.md) <br>\n- [Pharma Intelligence Workflow](references/pharma-intelligence-workflow.md) <br>\n- [Sources by Region](references/sources-by-region.md) <br>\n- [Regulatory Timelines by Agency](references/regulatory-timelines.md) <br>\n- [Drug Naming Conventions by Region](references/drug-naming.md) <br>\n- [SciMiner Unified MCP Endpoint](https://mcp.sciminer.tech/tools/unified/mcp) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [text, markdown, shell commands, guidance] <br>\n**Output Format:** [Markdown with structured tables, citations, and inline shell commands or HTTP request examples] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [Answers should cite source tier and access date, flag unresolved conflicts, and distinguish missing evidence from negative findings.] <br>\n\n## Skill Version(s): <br>\n1.0.1 (source: server release metadata) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.0.0: 7 files, 23988 bytes\n\nFiles: _meta.json (138b), references/drug-naming.md (4091b), references/mcp-tools.md (8738b), references/regulatory-timelines.md (5149b), references/sources-by-region.md (8513b), scripts/pharma-intelligence-workflow.md (9008b), SKILL.md (26074b)\n\nFile v1.0.0:SKILL.md\n\n---\nname: pharma-intelligence\ndescription: >\n  In-depth, multi-region pharmaceutical intelligence search and synthesis,\n  plus drug repurposing, target discovery, clinical evidence review, and\n  bioactivity analysis. Use this skill whenever the user asks about drug\n  approvals, clinical trials, regulatory submissions, pipeline assets, patent\n  landscapes, competitive intelligence, scientific evidence, disease targets,\n  genetic associations, or compound bioactivity for any drug, target,\n  indication, or company — especially when coverage of China, US, Europe,\n  Japan, South Korea, or Australia is needed. Trigger even for casual queries\n  like \"what's the approval status of X in China\", \"find trials for Y in\n  Japan\", \"compare pipeline coverage across regions\", \"find drugs for disease\n  Z\", or \"what targets are associated with condition W\". Always consult this\n  skill before answering any pharma or biomedical research question that\n  requires source-grounded data.\n---\n\n# Global Pharma Intelligence & Biomedical Research Skill\n\nSystematic, source-prioritized search and synthesis across regulatory, clinical,\nacademic, and commercial databases — covering all major pharmaceutical markets\nand 14+ biomedical research databases.\n\n## MCP Server\n\n**All structured queries go through:**\n```\nhttps://mcp.sciminer.tech/tools/unified/mcp\n```\nUse MCP tools before `web_search`/`web_fetch`. See **[references/mcp-tools.md](./references/mcp-tools.md)** for every tool's exact parameters and call format. For a complete worked example, see **[scripts/pharma-intelligence-workflow.md](./scripts/pharma-intelligence-workflow.md)**.\n\n## Quick Start — Drug Approval Status\n\n```json\n{\"jsonrpc\":\"2.0\",\"method\":\"tools/call\",\"params\":{\"name\":\"chembl_search_molecules\",\"arguments\":{\"query\":\"osimertinib\",\"max_results\":5}},\"id\":1}\n```\nThen check FDA label: `openfda_search_drug_labels` → `openfda_get_drug_label`.\nFor non-US regions (NMPA, EMA, PMDA): fall back to `web_fetch` (no MCP coverage — see Step 9 in workflow).\n\n---\n\n## Core Principle: Tiered Source Priority\n\nEvery region follows a **3-tier hierarchy**. Higher tiers override lower-tier\nclaims when there is conflict. Always cite the source tier in your response.\n\n| Tier | Type | Description |\n|------|------|-------------|\n| **Tier 1** | Regulatory | Official agency submissions, approvals, labels |\n| **Tier 2** | Trial registries | Prospective/registered clinical evidence |\n| **Tier 3** | Academic/IP | Published papers, conferences, patents |\n\n---\n\n## Region-by-Region Source Maps\n\n### 🇨🇳 China\n\n```\nTier 1 (Regulatory)\n  CDE       — drug evaluation, IND/NDA/sNDA submissions, clinical holds\n  NMPA      — final approvals, product licenses, post-market actions\n\nTier 2 (Trials)\n  chinadrugtrials.org.cn  — mandatory CN registry for IND-enab","readmeExcerpt":"Skill: Pharma Intelligence Owner: sciminer Summary: In-depth, multi-region pharmaceutical intelligence search and synthesis, plus drug repurposing, target discovery, clinical evidence review, and bioactivity analysis. Use this skill whenever the user asks about drug approvals, clinical trials, regulatory submissions, pipeline assets, patent landscape Tags: latest:1.0.3 Version history: v1.0.3 | 2026-07-06T15:14:37.49","codeSnippets":[],"executableExamples":[{"language":"bash","snippet":"echo '{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/biorxiv/2025-03-21/2025-03-28/0/json\",\"record_path\":\"collection\",\"max_items\":10}' | python scripts/rest_request.py"},{"language":"bash","snippet":"echo '{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/es_search/\",\"params\":{\"query\":\"caffeine\",\"size\":10},\"record_path\":\"results\",\"max_items\":10}' | python scripts/rest_request.py"},{"language":"bash","snippet":"echo '{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"activity.json\",\"params\":{\"molecule_chembl_id\":\"CHEMBL25\",\"limit\":10},\"record_path\":\"activities\",\"max_items\":10}' | python scripts/rest_request.py"},{"language":"bash","snippet":"echo '{\"action\":\"studies\",\"params\":{\"query.cond\":\"prostate cancer\",\"filter.overallStatus\":\"RECRUITING\",\"pageSize\":10},\"max_items\":10,\"max_pages\":1}' | python scripts/clinicaltrials_client.py"},{"language":"bash","snippet":"echo '{\"action\":\"search\",\"terms\":\"VCV000013080\",\"max_items\":10}' | python scripts/clinvar_variation.py"},{"language":"bash","snippet":"echo '{\"base_url\":\"https://www.ebi.ac.uk/ols4/api\",\"path\":\"search\",\"params\":{\"q\":\"asthma\",\"ontology\":\"efo\"},\"record_path\":\"response.docs\",\"max_items\":10}' | python scripts/rest_request.py"}],"parameters":null,"dependencies":[],"permissions":[],"extractedFiles":[{"path":"SKILL.md","content":"---\r\nname: pharma-intelligence\r\ndescription:\r\n  In-depth, multi-region pharmaceutical intelligence search and synthesis,\r\n  plus drug repurposing, target discovery, clinical evidence review, and\r\n  bioactivity analysis. Use this skill whenever the user asks about drug\r\n  approvals, clinical trials, regulatory submissions, pipeline assets, patent\r\n  landscapes, competitive intelligence, scientific evidence, disease targets,\r\n  genetic associations, or compound bioactivity for any drug, target,\r\n  indication, or company — especially when coverage of China, US, Europe,\r\n  Japan, South Korea, or Australia is needed. Trigger even for casual queries\r\n  like \"what's the approval status of X in China\", \"find trials for Y in\r\n  Japan\", \"compare pipeline coverage across regions\", \"find drugs for disease\r\n  Z\", or \"what targets are associated with condition W\". Always consult this\r\n  skill before answering any pharma or biomedical research question that\r\n  requires source-grounded data.\r\n---\r\n\r\n# Global Pharma Intelligence & Biomedical Research Skill\r\n\r\nSystematic, source-prioritized search and synthesis across regulatory, clinical,\r\nacademic, and commercial databases — covering all major pharmaceutical markets\r\nand 20+ biomedical research databases.\r\n\r\n## Sub-Skills — How to Invoke\r\n\r\nThis skill delegates all database work to the sub-skills bundled locally under `skills/`.\r\nRead the relevant sub-skill's `SKILL.md` before invoking it, then run its bundled script.\r\n\r\nSee [references/sub-skills.md](./references/sub-skills.md) for the full mapping of research\r\ntasks to sub-skills and execution patterns.\r\n\r\n---\r\n\r\n## Core Principle: Tiered Source Priority\r\n\r\nEvery region follows a 3-tier hierarchy. Higher tiers override lower-tier claims; always cite the tier.\r\n\r\n| Tier | Type | Description |\r\n|------|------|-------------|\r\n| **Tier 1** | Regulatory | Official agency submissions, approvals, labels |\r\n| **Tier 2** | Trial registries | Prospective/registered clinical evidence |\r\n| **Tier 3** | Academic / IP | Published papers, conferences, patents |\r\n\r\nFor the per-region source map (CN / US / EU / JP / KR / AU + global) with URLs and access notes, see [references/sources-by-region.md](./references/sources-by-region.md).\r\n\r\n---\r\n\r\n## Tool Access Notes\r\n\r\n`web_fetch` is the default tool for any URL in this skill that isn't covered by a bundled sub-skill. Some sites are JavaScript-rendered or block plain HTTP fetches — Google Patents is the most common offender, and CTIS, jRCT, and ANZCTR occasionally behave the same way — but this can happen on **any** site, not just those.\r\n\r\n**Rule:** try `web_fetch` first. If it returns empty, blocked, or placeholder content, retry the exact same URL with `browser_navigate` before concluding that a source has no data. Every other section in this skill that mentions `web_fetch` defers to this rule rather than restating it.\r\n\r\n---\r\n\r\n## Search Workflow\r\n\r\n### Step 1 — Classify the Query (pick ONE intent)\r\n\r\n| # | Intent | Trigger "},{"path":"skills/biorxiv-skill/SKILL.md","content":"---\r\nname: biorxiv-skill\r\ndescription: Submit compact bioRxiv and medRxiv API requests for details, publication-linkage, and DOI lookups. Use when a user wants concise preprint metadata summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/rest_request.py` for all bioRxiv and medRxiv API calls.\r\n- Use `base_url=https://api.biorxiv.org`.\r\n- The script accepts `max_items`; for `details` and `pubs` pages, start around `max_items=10`.\r\n- Prefer one cursor page at a time instead of increasing page size or pasting long collections into chat.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not part of the true request.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return the raw script JSON only if the user explicitly asks for machine-readable output.\r\n- Prefer these paths: `details/<server>/<start>/<end>/<cursor>/json`, `details/<server>/<doi>/na/json`, `pubs/<server>/<start>/<end>/<cursor>`, and `pubs/<server>/<doi>/na/json`.\r\n- If the user needs full page contents, set `save_raw=true` and report the saved file path rather than pasting large collections into chat.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required fields: `base_url`, `path`\r\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common biorxiv patterns:\r\n  - `{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/biorxiv/2025-03-21/2025-03-28/0/json\",\"record_path\":\"collection\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/medrxiv/10.1101/2020.09.09.20191205/na/json\",\"record_path\":\"collection\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"pubs/medrxiv/2020-03-01/2020-03-30/0\",\"record_path\":\"collection\",\"max_items\":10}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"base_url\":\"https://api.biorxiv.org\",\"path\":\"details/biorxiv/2025-03-21/2025-03-28/0/json\",\"record_path\":\"collection\",\"max_items\":10}' | python scripts/rest_request.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`."},{"path":"skills/chebi-skill/SKILL.md","content":"---\r\nname: chebi-skill\r\ndescription: Submit compact ChEBI 2.0 API requests for chemical search, compound lookup, ontology traversal, and structure metadata. Use when a user wants concise ChEBI summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/rest_request.py` for all ChEBI calls.\r\n- Use `base_url=https://www.ebi.ac.uk`.\r\n- Prefer the documented public routes under `chebi/backend/api/public/`.\r\n- Start with `es_search/` for free-text lookup and use `compound/<CHEBI:id>/` for targeted records.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return raw JSON only if the user explicitly asks for machine-readable output.\r\n- Prefer these paths: `chebi/backend/api/public/es_search/`, `chebi/backend/api/public/compound/<CHEBI:id>/`, and ontology child or parent routes.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required fields: `base_url`, `path`\r\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common ChEBI patterns:\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/es_search/\",\"params\":{\"query\":\"caffeine\",\"size\":10},\"record_path\":\"results\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/compound/CHEBI:27732/\"}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/ontology/children/CHEBI:27732/\"}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"base_url\":\"https://www.ebi.ac.uk\",\"path\":\"chebi/backend/api/public/es_search/\",\"params\":{\"query\":\"caffeine\",\"size\":10},\"record_path\":\"results\",\"max_items\":10}' | python scripts/rest_request.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`."},{"path":"skills/chembl-skill/SKILL.md","content":"---\r\nname: chembl-skill\r\ndescription: Submit compact ChEMBL API requests for activity, molecule, target, mechanism, and text-search endpoints. Use when a user wants concise ChEMBL summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/rest_request.py` for all ChEMBL API calls.\r\n- Use `base_url=https://www.ebi.ac.uk/chembl/api/data`.\r\n- The script accepts `max_items`; for activity, mechanism, and text-search collections, start with API `limit=10` and `max_items=10`.\r\n- Single molecule or target lookups usually do not need `max_items`.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Return the script JSON verbatim only if the user explicitly asks for machine-readable output.\r\n- Prefer these paths: `activity.json`, `molecule/<id>.json`, `target/<id>.json`, `mechanism.json`, and `molecule/search.json`.\r\n- Use `record_path` to target list fields like `activities`, `mechanisms`, or `molecules`.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required fields: `base_url`, `path`\r\n- Optional fields: `method`, `params`, `headers`, `json_body`, `form_body`, `record_path`, `response_format`, `max_items`, `max_depth`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common ChEMBL patterns:\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"activity.json\",\"params\":{\"molecule_chembl_id\":\"CHEMBL25\",\"limit\":10},\"record_path\":\"activities\",\"max_items\":10}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"molecule/CHEMBL25.json\"}`\r\n  - `{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"molecule/search.json\",\"params\":{\"q\":\"imatinib\",\"limit\":10},\"record_path\":\"molecules\",\"max_items\":10}`\r\n\r\n## Output\r\n- Success returns `ok`, `source`, `path`, `method`, `status_code`, `warnings`, and either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"base_url\":\"https://www.ebi.ac.uk/chembl/api/data\",\"path\":\"activity.json\",\"params\":{\"molecule_chembl_id\":\"CHEMBL25\",\"limit\":10},\"record_path\":\"activities\",\"max_items\":10}' | python scripts/rest_request.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/rest_request.py`."},{"path":"skills/clinicaltrials-skill/SKILL.md","content":"---\r\nname: clinicaltrials-skill\r\ndescription: Submit compact ClinicalTrials.gov API v2 requests for study search, metadata, enums, search areas, and field statistics. Use when a user wants concise ClinicalTrials.gov summaries\r\n---\r\n\r\n## Operating rules\r\n- Use `scripts/clinicaltrials_client.py` for all ClinicalTrials.gov v2 calls.\r\n- Study searches are better with `max_items=10` and `max_pages=1`; only increase pages when the user explicitly wants more than the first page.\r\n- Use targeted `params` instead of broad unfiltered study dumps.\r\n- Re-run requests in long conversations instead of relying on older tool output.\r\n- Treat displayed `...` in tool previews as UI truncation, not literal request content.\r\n\r\n## Execution behavior\r\n- Return concise markdown summaries from the script JSON by default.\r\n- Prefer `action=studies` for search and `action=metadata|search_areas|enums|stats_size|field_values|field_sizes` for API introspection and field stats.\r\n- If the user needs full pages or aggregated responses, set `save_raw=true` and report the saved file path.\r\n\r\n## Input\r\n- Read one JSON object from stdin.\r\n- Required field: `action`\r\n- Supported actions: `studies`, `metadata`, `search_areas`, `enums`, `stats_size`, `field_values`, `field_sizes`, `request`\r\n- Optional fields: `path` for `action=request`, `params`, `max_items`, `max_depth`, `max_pages`, `timeout_sec`, `save_raw`, `raw_output_path`\r\n- Common ClinicalTrials.gov patterns:\r\n  - `{\"action\":\"studies\",\"params\":{\"query.cond\":\"prostate cancer\",\"filter.overallStatus\":\"RECRUITING\",\"pageSize\":10},\"max_items\":10,\"max_pages\":1}`\r\n  - `{\"action\":\"metadata\"}`\r\n  - `{\"action\":\"field_values\",\"params\":{\"field\":\"protocolSection.identificationModule.organization.fullName\"}}`\r\n\r\n## Output\r\n- `action=studies` returns `pages_fetched`, `next_page_token`, count metadata, and compact `records`.\r\n- Other actions return either compact `records` or a compact `summary`.\r\n- Use `raw_output_path` when `save_raw=true`.\r\n- Failure returns `ok=false` with `error.code` and `error.message`.\r\n\r\n## Execution\r\n```bash\r\necho '{\"action\":\"studies\",\"params\":{\"query.cond\":\"prostate cancer\",\"filter.overallStatus\":\"RECRUITING\",\"pageSize\":10},\"max_items\":10,\"max_pages\":1}' | python scripts/clinicaltrials_client.py\r\n```\r\n\r\n## References\r\n- No additional runtime references are required; keep the import package limited to this file and `scripts/clinicaltrials_client.py`."}],"languages":[],"docsSourceLabel":"CLAWHUB","editorialOverview":null,"editorialQuality":{"score":100,"threshold":65,"status":"thin","wordCount":2413,"uniquenessScore":33,"reasons":["uniqueness-below-45"]}},"media":{"evidence":{"source":"no-media","verified":false,"confidence":"low","updatedAt":"2026-10-10T23:56:50.891Z","emptyReason":"No screenshots, media assets, or demo links are available."},"primaryImageUrl":null,"mediaAssetCount":0,"assets":[],"demoUrl":null},"ownerResources":{"evidence":{"source":"unclaimed","verified":false,"confidence":"low","updatedAt":"2026-10-10T23:56:50.891Z","emptyReason":"This page has not been claimed by the agent owner."},"hasCustomPage":false,"customPageUpdatedAt":null,"customLinks":[],"structuredLinks":{"docsUrl":null,"demoUrl":null,"supportUrl":null,"pricingUrl":null,"statusUrl":null},"customPage":null},"relatedAgents":{"evidence":{"source":"protocol-neighbors","verified":false,"confidence":"medium","updatedAt":"2026-10-11T03:54:16.182Z","emptyReason":null},"items":[{"id":"8ebccd8e-3863-4187-8355-c3f14e1f9edf","entityType":"agent","canonicalPath":"/agent/iofficeai-aionui","slug":"iofficeai-aionui","name":"AionUi","description":"Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!","url":"https://github.com/iOfficeAI/AionUi","homepage":"https://www.aionui.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-10-09T19:11:12.944Z","createdAt":"2026-02-25T03:38:16.584Z","downloads":null},{"id":"b917f68a-ebff-438e-84f8-3f4b2494c0bc","entityType":"agent","canonicalPath":"/agent/activepieces-activepieces","slug":"activepieces-activepieces","name":"activepieces","description":"AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents","url":"https://github.com/activepieces/activepieces","homepage":"https://www.activepieces.com","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-15T02:22:12.426Z","createdAt":"2026-02-25T03:38:12.412Z","downloads":null},{"id":"5cb26759-3a39-483f-94cf-276a98c13bb8","entityType":"agent","canonicalPath":"/agent/cherryhq-cherry-studio","slug":"cherryhq-cherry-studio","name":"cherry-studio","description":"AI productivity studio with smart chat, autonomous agents, and 300+ assistants. Unified access to frontier LLMs","url":"https://github.com/CherryHQ/cherry-studio","homepage":"https://cherry-ai.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-11T14:38:40.986Z","createdAt":"2026-02-25T03:38:19.379Z","downloads":null},{"id":"6f6582d0-5d76-4f0f-b81d-86520247950b","entityType":"agent","canonicalPath":"/agent/copilotkit-copilotkit","slug":"copilotkit-copilotkit","name":"CopilotKit","description":"The Frontend for Agents & Generative UI. React + Angular","url":"https://github.com/CopilotKit/CopilotKit","homepage":"https://docs.copilotkit.ai","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-03-25T09:50:57.846Z","createdAt":"2026-02-25T03:39:14.617Z","downloads":null}],"links":{"hub":"/agent","source":"/agent/source/clawhub","protocols":[{"label":"OpenClaw","href":"/agent/protocol/openclew"}]}}}