{"id":"e6880f9a-b4a7-4898-812f-68d932e28320","entityType":"agent","slug":"clawhub-leostehlik-brief-master","name":"Brief Master","canonicalUrl":"https://www.xpersona.co/agent/clawhub-leostehlik-brief-master","canonicalPath":"/agent/clawhub-leostehlik-brief-master","generatedAt":"2026-10-12T16:27:00.989Z","source":"CLAWHUB","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T22:47:00.838Z","emptyReason":null},"description":"Write precise execution briefs for agents, cron jobs, reviewers, researchers, and delegated coding tasks.","descriptionLabel":"Source description","evidenceSummary":"Capability contract not published. No trust telemetry is available yet. 982 downloads reported by the source. Last updated 10/11/2026.","installCommand":"clawhub skill install s1754kcncc002avbhpwjnqgcr98728et:brief-master","sourceUrl":"https://clawhub.ai/leostehlik/brief-master","homepage":"https://clawhub.ai/leostehlik/skills/brief-master","primaryLinks":[{"label":"View on ClawHub","url":"https://clawhub.ai/leostehlik/brief-master","kind":"source"},{"label":"Homepage","url":"https://clawhub.ai/leostehlik/skills/brief-master","kind":"homepage"}],"safetyScore":84,"overallRank":62,"popularityScore":60,"trustScore":null,"claimedByName":null,"isOwner":false,"seoDescription":"Brief Master technical dossier on Xpersona with agent coverage, OPENCLEW support, and live trust metadata."},"coverage":{"evidence":{"source":"public-profile","verified":false,"confidence":"medium","updatedAt":"2026-10-11T22:47:00.838Z","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":"2026-10-11T22:47:00.838Z","emptyReason":null},"stars":null,"forks":null,"downloads":982,"likes":null,"task":null,"library":null,"packageName":null,"latestVersion":"0.2.1","tractionLabel":"982 downloads"},"release":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T22:47:00.704Z","emptyReason":null},"lastUpdatedAt":"2026-10-11T22:47:00.838Z","lastCrawledAt":"2026-10-11T22:47:00.704Z","lastIndexedAt":null,"nextCrawlAt":"2026-10-12T22:47:00.704Z","lastVerifiedAt":null,"highlights":[{"version":"0.2.1","createdAt":"2026-05-24T10:00:02.661Z","changelog":"Document that Brief Master drafts briefs only, requires review before autonomous execution, and must not invent credentials, approvals, hosts, cron schedules, or elevated access.","fileCount":10,"zipByteSize":11703},{"version":"0.2.0","createdAt":"2026-05-22T17:05:05.301Z","changelog":"Shorten metadata, scrub private templates, add public-safe brief formats, and add version metadata.","fileCount":6,"zipByteSize":6716},{"version":"0.1.0","createdAt":"2026-05-20T23:41:56.714Z","changelog":"Initial public Brief Master skill with public-safe templates.","fileCount":6,"zipByteSize":7988}]},"execution":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No published capability contract is available yet."},"installCommand":"clawhub skill install s1754kcncc002avbhpwjnqgcr98728et:brief-master","setupComplexity":"low","setupSteps":["Install using `clawhub skill install s1754kcncc002avbhpwjnqgcr98728et:brief-master` 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/leostehlik/brief-master 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-leostehlik-brief-master/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-leostehlik-brief-master/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-leostehlik-brief-master/trust"},"curlExamples":["curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-leostehlik-brief-master/snapshot\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-leostehlik-brief-master/contract\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-leostehlik-brief-master/trust\""],"jsonRequestTemplate":{"query":"summarize this repo","constraints":{"maxLatencyMs":2000,"protocolPreference":["OPENCLEW"]}},"jsonResponseTemplate":{"ok":true,"result":{"summary":"...","confidence":0.9},"meta":{"source":"CLAWHUB","generatedAt":"2026-10-12T16:27:00.988Z"}},"retryPolicy":{"maxAttempts":3,"backoffMs":[500,1500,3500],"retryableConditions":["HTTP_429","HTTP_503","NETWORK_TIMEOUT"]}},"endpoints":{"dossierUrl":"https://www.xpersona.co/api/v1/agents/clawhub-leostehlik-brief-master/dossier","snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-leostehlik-brief-master/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-leostehlik-brief-master/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-leostehlik-brief-master/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":"2026-10-11T22:47:00.838Z","emptyReason":null},"readme":"Skill: Brief Master\n\nOwner: leostehlik\n\nSummary: Write precise execution briefs for agents, cron jobs, reviewers, researchers, and delegated coding tasks.\n\nTags: agents:0.2.1, briefs:0.2.1, latest:0.2.1, openclaw:0.2.1, prompting:0.2.1\n\nVersion history:\n\nv0.2.1 | 2026-05-24T10:00:02.661Z | user\n\nDocument that Brief Master drafts briefs only, requires review before autonomous execution, and must not invent credentials, approvals, hosts, cron schedules, or elevated access.\n\nv0.2.0 | 2026-05-22T17:05:05.301Z | user\n\nShorten metadata, scrub private templates, add public-safe brief formats, and add version metadata.\n\nv0.1.0 | 2026-05-20T23:41:56.714Z | user\n\nInitial public Brief Master skill with public-safe templates.\n\nArchive index:\n\nArchive v0.2.1: 10 files, 11703 bytes\n\nFiles: examples/messy-to-clean-brief.md (1882b), examples/README.md (420b), LICENSE (1068b), README.md (5467b), references/9-dimensions.md (2338b), references/brief-formats.md (3062b), scripts/validate_skill_repo.py (2232b), skill-card.md (2039b), SKILL.md (2854b), _meta.json (131b)\n\nFile v0.2.1:SKILL.md\n\n---\nname: brief-master\ndescription: \"Write precise execution briefs for agents, cron jobs, reviewers, researchers, and delegated coding tasks.\"\nmetadata:\n  version: \"0.2.1\"\n---\n# Brief Master\n\nWrite agent briefs that agents actually execute correctly.\n\nEvery wasted token is a wasted API call. Every vague word is a future bug. Every missing acceptance criterion is a future rework cycle.\n\nRead `references/9-dimensions.md` before extracting intent.\nRead `references/brief-formats.md` for the right format per agent type.\n\n\n## Safety Boundary\n\nThis skill writes briefs only. A generated brief can instruct another agent to edit code, run commands, schedule jobs, or use external services, so the user or orchestrator must review the final brief before execution.\n\nDo not invent credentials, secret names, hostnames, destructive commands, cron schedules, repository permissions, or approval status. If a brief would require elevated access, remote execution, persistent automation, or public posting, mark that as requiring explicit approval in the brief.\n\n## The Pipeline\n\n1. **Detect the target** — which agent, what runtime (subagent, cron, sessions_spawn)?\n2. **Extract 9 dimensions** — see `references/9-dimensions.md`\n3. **Ask max 3 questions** — only if critical info is missing. Never more.\n4. **Apply the right format** — see `references/brief-formats.md`\n5. **Run token efficiency audit** — strip every word that doesn't change the output\n6. **Deliver** — one clean brief, ready to use\n\n## Token Efficiency Rule\n\n\"The best brief is not the longest. It's the one where every word is load-bearing.\"\n\nBefore delivering, ask: does removing this sentence change what the agent does? If not, cut it.\n\n## What Makes a Bad Brief\n\n- Vague task description (\"improve the translation\")\n- No acceptance criteria (how does the agent know it's done?)\n- Missing constraints (what must not break?)\n- No non-goals (what's explicitly out of scope?)\n- Too long (agent loses focus, context drifts)\n- Tool-specific instructions missing (which host? which directory? which branch?)\n\n## What Makes a Good Brief\n\n- One clear task per brief\n- Explicit ACs labelled AC1, AC2, AC3 (testable, not descriptions)\n- Constraints listed\n- Non-goals listed\n- Correct status update blocks (start + end)\n- 95% confidence gate instruction included\n- Right tool commands with correct paths\n\n## Mandatory Sections for Dev Agent Briefs\n\nEvery brief for a dev/build agent must include:\n\n```\nFIRST — update status to working with task description\n...task...\nLAST — update status to done + notify orchestrator\n```\n\nAnd the 95% confidence gate:\n```\nBefore starting ANY work, ask clarifying questions until you are 95% confident\nyou can complete this task successfully. Do not start until you have that confidence.\n```\n\nSee `references/brief-formats.md` for full templates.\n\nFile v0.2.1:examples/README.md\n\n# Brief Master Examples\n\nStart with [`messy-to-clean-brief.md`](messy-to-clean-brief.md).\n\nIt shows the core move of Brief Master: take a fuzzy request, preserve the intent, and turn it into an agent brief with scope, constraints, acceptance criteria, non-goals, and verification commands.\n\nUse it when you want to see what changes between \"please improve this\" and a brief an implementation agent can actually execute.\n\nFile v0.2.1:README.md\n\n# brief-master\n\n![Validate](https://github.com/LeoStehlik/brief-master/actions/workflows/validate.yml/badge.svg)\n\nWrite sharp, precise agent briefs for OpenClaw. Zero wasted tokens. Zero vague instructions.\n\nBrief Master is an agent brief writer for AI coding workflows. It extracts 9 dimensions of intent, asks max 3 clarifying questions, applies the right format for your target agent, runs a token efficiency audit, and delivers one clean brief ready to fire.\n\nBuilt for OpenClaw's multi-agent workflow: dev agents, code reviewers, testers, researchers, and cron jobs.\n\nInspired by [prompt-master](https://github.com/nidhinjs/prompt-master). Built as our own OpenClaw-native implementation.\n\n\n## Safety Boundary\n\nBrief Master only drafts execution briefs. Review the generated brief before handing it to an autonomous agent, especially commands, file paths, credentials, scheduled jobs, external services, and anything that changes code or system state.\n\nThe skill must not invent access, approval, hostnames, secrets, destructive commands, or cron schedules. If execution needs elevated privileges or persistent automation, the brief should say that explicit approval is required.\n\n## Use Cases\n\n- turn a messy request into a precise agent brief\n- make acceptance criteria explicit before an agent starts work\n- capture constraints, non-goals, inputs, and verification steps\n- reduce rework from vague prompts and missing context\n\n---\n\n## The Problem\n\nEvery wasted token is a wasted API call.\nEvery vague word is a future bug.\nEvery missing acceptance criterion is a future rework cycle.\n\n\"Improve the translation system\" is not a brief.\n\"Fix formSchemaTranslated being NULL for German locale — AC1: all 50 prompt rows have formSchemaTranslated populated after running translate/de endpoint\" is a brief.\n\n---\n\n## Credibility Artifact\n\nSee [`examples/README.md`](examples/README.md) for the example index, or jump straight to [`examples/messy-to-clean-brief.md`](examples/messy-to-clean-brief.md) for a concrete messy request -> clean agent brief example with ACs, constraints, non-goals, and verification commands.\n\n## The Pipeline\n\n1. Detect the target agent and runtime\n2. Extract 9 dimensions of intent\n3. Ask max 3 clarifying questions (only if critical info is missing)\n4. Apply the right format for the target agent\n5. Run token efficiency audit\n6. Deliver one clean, ready-to-use brief\n\n---\n\n## The 9 Dimensions\n\n1. **Task** - what exactly should be done (one verb, one outcome)\n2. **Input** - what does the agent need to start (files, machines, branches)\n3. **Output** - what should exist when done\n4. **Constraints** - what must not break\n5. **Context** - only what changes what the agent does\n6. **Audience/Role** - which agent, what runtime\n7. **Memory** - what files to read first\n8. **Success Criteria** - AC1, AC2, AC3 (testable, specific)\n9. **Examples** - reference points that clarify expected behaviour\n\n---\n\n\n## When To Use Which Repo\n\nUse this repo when the task is still too vague to hand to an agent safely. Brief Master turns messy intent into a clear brief with task scope, inputs, outputs, constraints, non-goals, acceptance criteria, and verification shape.\n\nUse the neighbouring tools at different points in the workflow:\n\n| Need | Use |\n| --- | --- |\n| Turn a fuzzy request into an executable agent brief | [Brief Master](https://github.com/LeoStehlik/brief-master) |\n| Prove one coding task is actually done | [Proof Loop](https://github.com/LeoStehlik/proof-loop) |\n| Improve repeated agent behaviour with evals | [Loopsmith](https://github.com/LeoStehlik/loopsmith) |\n| Keep source-backed memory for long-running agents | [Sovereign Brain](https://github.com/LeoStehlik/decoupled-agent-memory) |\n| Stop frontend agents producing generic UI sludge | [no-slop-ui](https://github.com/LeoStehlik/no-slop-ui) |\n\nA practical chain looks like this: messy request -> Brief Master brief -> Proof Loop task -> Loopsmith eval if the same failure keeps recurring -> Sovereign Brain records the durable decision.\n\n## Related Tools\n\n- [Proof Loop](https://github.com/LeoStehlik/proof-loop) - use after Brief Master when the task needs frozen acceptance criteria, separate verifier roles, and durable proof artifacts.\n- [Loopsmith](https://github.com/LeoStehlik/loopsmith) - use when recurring weak briefs or agent failures should become eval cases and promotion decisions.\n- [Sovereign Brain](https://github.com/LeoStehlik/decoupled-agent-memory) - use as source-backed context for briefs that need current decisions, project state, or changed evidence.\n\n## Installation\n\n### OpenClaw\n\n```json\n{\n  \"skills\": {\n    \"load\": {\n      \"extraDirs\": [\"/path/to/your/skills\"]\n    }\n  }\n}\n```\n\n```bash\ngit clone https://github.com/LeoStehlik/brief-master.git /path/to/your/skills/brief-master\n```\n\n### Claude Code / Codex\n\nCopy into `.agents/skills/brief-master/` or `.claude/skills/brief-master/`.\n\n---\n\n## Usage\n\n```\nWrite me a brief for my dev agent to fix the translation timeout issue\n```\n\n```\nHelp me write a cron job prompt for overnight research\n```\n\n```\nI need a code review brief for Sprint 5\n```\n\n---\n\n## What's Inside\n\n```\nbrief-master/\n├── SKILL.md                           Core pipeline + quality rules\n└── references/\n    ├── 9-dimensions.md                Intent extraction framework\n    └── brief-formats.md               Ready-to-use templates per agent type\n```\n\n---\n\n## License\n\nMIT - see [LICENSE](LICENSE)\n\nFile v0.2.1:_meta.json\n\n{\n  \"ownerId\": \"kn7d3r58cdxk8k0xg6jq6k65gs873c0c\",\n  \"slug\": \"brief-master\",\n  \"version\": \"0.2.1\",\n  \"publishedAt\": 1779616802661\n}\n\nFile v0.2.1:references/9-dimensions.md\n\n# 9 Dimensions of Intent\n\nExtract only the dimensions needed to make the brief executable. Do not turn this into an interview script when context is already clear.\n\n## 1. Task\n\nWhat should the agent do?\n\n- One verb, one outcome\n- Split unrelated tasks into separate briefs\n- Prefer \"fix X\" / \"implement Y\" / \"review Z\" over vague \"work on\"\n\n## 2. Input\n\nWhat does the agent need to start?\n\n- Repo, files, URLs, datasets, credentials, or screenshots\n- Runtime/host if relevant\n- Branch, directory, service, or environment\n\nAvoid vague location text like \"in the repo\" when the exact path or branch matters.\n\n## 3. Output\n\nWhat should exist when the agent is done?\n\n- Files created or changed\n- Tests/proof artifacts generated\n- Commit, PR, report, or message expected\n- Service or data state expected\n\n## 4. Constraints\n\nWhat must not break?\n\n- Existing behaviours\n- Files, data, credentials, or generated assets that must not be touched\n- Backwards compatibility or performance limits\n- Security/privacy boundaries\n\n## 5. Context\n\nWhat does the agent need to understand about why this task exists?\n\nKeep it short. Include only context that changes the work.\n\n## 6. Audience / Role\n\nWho is the brief for?\n\n- Builder / implementer\n- Reviewer / verifier\n- Researcher\n- Cron or scheduled agent\n- Human operator\n\nRole determines autonomy, tool expectations, and output shape.\n\n## 7. Prior Context\n\nWhat should the agent read first?\n\n- Specs, issue links, previous verdicts, relevant docs\n- Existing implementation patterns\n- Prior decisions or non-goals\n\n## 8. Success Criteria\n\nHow does everyone know it is done?\n\n- Use explicit AC1, AC2, AC3 where possible\n- Each AC must be testable by a fresh verifier\n- Include the exact verification command or manual check when known\n\nIf you cannot write a testable AC, the task is under-defined.\n\n## 9. Examples\n\nWhat examples clarify the intended result?\n\n- Before/after screenshots\n- Expected output format\n- Similar accepted implementation\n- Known-good command output\n\n## Quick Extraction Prompt\n\nAsk at most three questions in one round:\n\n```text\nI can write the brief, but I need three specifics:\n1. [missing blocker]\n2. [missing blocker]\n3. [missing blocker]\n```\n\nIf more than three answers are missing, ask the three that most affect correctness and note the remaining assumptions in the brief.\n\nFile v0.2.1:references/brief-formats.md\n\n# Brief Formats\n\nUse these as public-safe templates. Replace placeholders with the user's actual environment only when the user has provided it or it is already part of the current task context.\n\n## Builder Brief\n\n```markdown\n# Brief: [TASK_ID] - [one-line outcome]\n\n## Confidence Gate\n\nBefore starting implementation, ask clarifying questions until you are 95% confident you can complete this task correctly. If you are already 95% confident, state assumptions and proceed.\n\n## Task\n\n[One clear sentence describing what to build or fix.]\n\n## Context\n\n[Only context that changes the implementation. 2-5 sentences.]\n\n## Acceptance Criteria\n\nAC1: [specific, testable condition]\nVerify: [command/check]\n\nAC2: [specific, testable condition]\nVerify: [command/check]\n\nAC3: [specific, testable condition]\nVerify: [command/check]\n\n## Constraints\n\n- [What must not break]\n- [Files/data/services that must not be touched]\n- [Security/privacy boundaries]\n\n## Non-Goals\n\n- [Explicitly out of scope]\n\n## Read First\n\n- [file/doc/spec/issue link]\n\n## Environment\n\n- Repo/path: [repo/path]\n- Branch: [branch]\n- Runtime/service: [if relevant]\n\n## Deliver\n\n- Changed files\n- Verification output\n- Known residual risks\n```\n\n## Reviewer / Verifier Brief\n\n```markdown\n# Review: [TASK_ID]\n\n## Role\n\nYou are the verifier. Do not implement fixes. Verify the acceptance criteria from a fresh session and report evidence.\n\n## Read First\n\n- [spec or brief]\n- [existing verdict/problems if any]\n\n## Acceptance Criteria To Verify\n\nAC1: [copy from spec]\nAC2: [copy from spec]\nAC3: [copy from spec]\n\n## Required Checks\n\n- [test command]\n- [manual/browser/API check]\n- [regression check]\n\n## Output\n\nReturn:\n- PASS/FAIL/UNKNOWN for each AC\n- Evidence for each verdict\n- File/line references for failures when applicable\n- Test commands run and exact result\n```\n\n## Research Brief\n\n```markdown\n# Research: [topic]\n\n## Question\n\n[Specific question to answer.]\n\n## Scope\n\n- Include: [source types, date range, domains, handles]\n- Exclude: [known noise]\n\n## Source Quality\n\n- Prefer primary sources.\n- Include URLs and publication dates.\n- Separate observed facts from inference.\n- Flag uncertainty.\n\n## Output\n\n- Short answer\n- Evidence table\n- Implications\n- Open questions\n```\n\n## Cron / Scheduled Agent Brief\n\n```markdown\n# Scheduled Task: [name]\n\n## Goal\n\n[What should be checked or produced.]\n\n## Steps\n\n1. [Exact action]\n2. [Exact action]\n3. [Exact action]\n\n## Success Condition\n\n[What normal healthy output means.]\n\n## Alert Condition\n\nSend a message only if:\n- [condition]\n- [condition]\n\nIf everything is healthy, return `NO_REPLY`.\n\n## Output\n\n- Write report to [path] if needed.\n- Include exact failure evidence if alerting.\n```\n\n## Token Efficiency Checklist\n\n- Can any sentence be removed without changing what the agent does?\n- Are paths, branches, commands, and URLs specific where needed?\n- Are ACs testable by someone who did not write the code?\n- Are constraints and non-goals explicit?\n- Are private local assumptions excluded unless required?\n- Is the final output shape clear?\n\nFile v0.2.1:examples/messy-to-clean-brief.md\n\n# Messy Request To Clean Brief\n\nBrief Master exists for the moment before Proof Loop: the user knows what hurts, but the task is still too vague to hand to an agent safely.\n\n## Messy Request\n\n```text\nThe German translation thing is still broken. Can you get an agent to fix it properly? It timed out last time and I do not want another half-fix.\n```\n\n## Clean Agent Brief\n\n```markdown\n## Task\nFix German prompt translation persistence so translated form schemas are populated reliably.\n\n## Inputs\n- Repo: `/workspace/app`\n- Start with `apps/server/src/prompts/translation.service.ts`\n- Read the existing translation timeout issue and latest failing test output first.\n\n## Acceptance Criteria\nAC1: All 50 German prompt rows have `formSchemaTranslated` populated after running the translation endpoint.\nVerify: run `pnpm test:server translation` and inspect the seeded German prompt rows.\n\nAC2: The translation job handles provider timeout without leaving partial NULL state.\nVerify: use the existing timeout fixture or add one focused test if missing.\n\nAC3: Existing English prompt behaviour is unchanged.\nVerify: run the current prompt service tests.\n\n## Constraints\n- Do not change the public API response shape.\n- Do not rewrite unrelated prompt storage code.\n- Keep the fix small enough for focused review.\n\n## Non-goals\n- No new translation provider.\n- No UI work.\n- No broad prompt model migration.\n\n## Verification\nRun the focused server tests first. If they pass, run the broader prompt test suite. Record command output in the task evidence.\n```\n\n## What Changed\n\nThe cleaned brief turns a complaint into something an agent can execute:\n\n- one task\n- known starting files\n- explicit acceptance criteria\n- constraints and non-goals\n- verification commands\n- evidence expectations\n\nThat is the handoff point where Proof Loop can freeze the spec and make completion auditable.\n\nFile v0.2.1:skill-card.md\n\n## Description:\n\nWrite precise execution briefs for agents, cron jobs, reviewers, researchers, and delegated coding tasks.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[leostehlik](https://clawhub.ai/user/leostehlik)\n\n### License/Terms of Use:\n\nMIT\n\n## Use Case:\n\nDevelopers and AI workflow operators use Brief Master to turn vague requests into execution briefs for agents, reviewers, researchers, and scheduled jobs, with explicit acceptance criteria, constraints, non-goals, and verification steps.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: Generated briefs may include commands, schedules, external services, credentials, or changes to code or data that could be unsafe if handed directly to an autonomous agent.\n\nMitigation: Review generated briefs before execution and require explicit approval for elevated access, remote execution, persistent automation, public posting, credentials, destructive commands, hostnames, or cron schedules.\n\n## Reference(s):\n\n- [9 Dimensions of Intent](references/9-dimensions.md)\n- [Brief Formats](references/brief-formats.md)\n- [Messy Request To Clean Brief](examples/messy-to-clean-brief.md)\n- [Brief Master on ClawHub](https://clawhub.ai/leostehlik/skills/brief-master)\n\n## Skill Output:\n\n**Output Type(s):** [text, markdown, shell commands, configuration, guidance]\n\n**Output Format:** [Markdown brief with labeled sections, acceptance criteria, constraints, non-goals, and inline verification commands when supplied by the user.]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [Produces one concise execution brief for review before autonomous use.]\n\n## Skill Version(s):\n\n0.2.1 (source: SKILL.md frontmatter and server release metadata)\n\n## Ethical Considerations:\n\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.\n\nFile v0.2.1:LICENSE\n\nMIT License\n\nCopyright (c) 2026 Leo Stehlik\n\nPermission is hereby granted, free of charge, to any person obtaining a copy\nof this software and associated documentation files (the \"Software\"), to deal\nin the Software without restriction, including without limitation the rights\nto use, copy, modify, merge, publish, distribute, sublicense, and/or sell\ncopies of the Software, and to permit persons to whom the Software is\nfurnished to do so, subject to the following conditions:\n\nThe above copyright notice and this permission notice shall be included in all\ncopies or substantial portions of the Software.\n\nTHE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\nIMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\nFITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE\nAUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\nLIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,\nOUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE\nSOFTWARE.\n\nArchive v0.2.0: 6 files, 6716 bytes\n\nFiles: LICENSE (1068b), README.md (2667b), references/9-dimensions.md (2338b), references/brief-formats.md (3062b), SKILL.md (2504b), _meta.json (131b)\n\nFile v0.2.0:SKILL.md\n\n---\nname: brief-master\ndescription: \"Write precise execution briefs for agents, cron jobs, reviewers, researchers, and delegated coding tasks.\"\nmetadata:\n  version: \"0.2.0\"\n---\n\n# Brief Master\n\nWrite agent briefs that agents actually execute correctly.\n\nEvery wasted token is a wasted API call. Every vague word is a future bug. Every missing acceptance criterion is a future rework cycle.\n\nRead `references/9-dimensions.md` before extracting intent.\nRead `references/brief-formats.md` for the right format per agent type.\n\nDo not include stale local agent names, private host paths, dashboard ceremony, chat IDs, or workspace-specific status updates unless the user explicitly asks for that environment.\n\n## The Pipeline\n\n1. **Detect the target** — which agent, what runtime (subagent, cron, sessions_spawn)?\n2. **Extract 9 dimensions** — see `references/9-dimensions.md`\n3. **Ask max 3 questions** — only if critical info is missing. Never more.\n4. **Apply the right format** — see `references/brief-formats.md`\n5. **Run token efficiency audit** — strip every word that doesn't change the output\n6. **Deliver** — one clean brief, ready to use\n\n## Token Efficiency Rule\n\n\"The best brief is not the longest. It's the one where every word is load-bearing.\"\n\nBefore delivering, ask: does removing this sentence change what the agent does? If not, cut it.\n\n## What Makes a Bad Brief\n\n- Vague task description (\"improve the translation\")\n- No acceptance criteria (how does the agent know it's done?)\n- Missing constraints (what must not break?)\n- No non-goals (what's explicitly out of scope?)\n- Too long (agent loses focus, context drifts)\n- Tool-specific instructions missing (which host? which directory? which branch?)\n\n## What Makes a Good Brief\n\n- One clear task per brief\n- Explicit ACs labelled AC1, AC2, AC3 (testable, not descriptions)\n- Constraints listed\n- Non-goals listed\n- Correct status update blocks (start + end)\n- 95% confidence gate instruction included\n- Right tool commands with correct paths\n\n## Mandatory Sections for Dev Agent Briefs\n\nEvery brief for a dev/build agent must include:\n\n```\nFIRST — update status to working with task description\n...task...\nLAST — update status to done + notify orchestrator\n```\n\nAnd the 95% confidence gate:\n```\nBefore starting ANY work, ask clarifying questions until you are 95% confident\nyou can complete this task successfully. Do not start until you have that confidence.\n```\n\nSee `references/brief-formats.md` for full templates.\n\nFile v0.2.0:README.md\n\n# brief-master\n\nWrite sharp, precise agent briefs for OpenClaw. Zero wasted tokens. Zero vague instructions.\n\nEvery bad brief costs you a rework cycle. brief-master extracts 9 dimensions of intent, asks max 3 clarifying questions, applies the right format for your target agent, runs a token efficiency audit, and delivers one clean brief ready to fire.\n\nBuilt for OpenClaw's multi-agent workflow: dev agents, code reviewers, testers, researchers, and cron jobs.\n\nInspired by [prompt-master](https://github.com/nidhinjs/prompt-master). Built as our own OpenClaw-native implementation.\n\n---\n\n## The Problem\n\nEvery wasted token is a wasted API call.\nEvery vague word is a future bug.\nEvery missing acceptance criterion is a future rework cycle.\n\n\"Improve the translation system\" is not a brief.\n\"Fix formSchemaTranslated being NULL for German locale — AC1: all 50 prompt rows have formSchemaTranslated populated after running translate/de endpoint\" is a brief.\n\n---\n\n## The Pipeline\n\n1. Detect the target agent and runtime\n2. Extract 9 dimensions of intent\n3. Ask max 3 clarifying questions (only if critical info is missing)\n4. Apply the right format for the target agent\n5. Run token efficiency audit\n6. Deliver one clean, ready-to-use brief\n\n---\n\n## The 9 Dimensions\n\n1. **Task** - what exactly should be done (one verb, one outcome)\n2. **Input** - what does the agent need to start (files, machines, branches)\n3. **Output** - what should exist when done\n4. **Constraints** - what must not break\n5. **Context** - only what changes what the agent does\n6. **Audience/Role** - which agent, what runtime\n7. **Memory** - what files to read first\n8. **Success Criteria** - AC1, AC2, AC3 (testable, specific)\n9. **Examples** - reference points that clarify expected behaviour\n\n---\n\n## Installation\n\n### OpenClaw\n\n```json\n{\n  \"skills\": {\n    \"load\": {\n      \"extraDirs\": [\"/path/to/your/skills\"]\n    }\n  }\n}\n```\n\n```bash\ngit clone https://github.com/LeoStehlik/brief-master.git /path/to/your/skills/brief-master\n```\n\n### Claude Code / Codex\n\nCopy into `.agents/skills/brief-master/` or `.claude/skills/brief-master/`.\n\n---\n\n## Usage\n\n```\nWrite me a brief for my dev agent to fix the translation timeout issue\n```\n\n```\nHelp me write a cron job prompt for overnight research\n```\n\n```\nI need a code review brief for Sprint 5\n```\n\n---\n\n## What's Inside\n\n```\nbrief-master/\n├── SKILL.md                           Core pipeline + quality rules\n└── references/\n    ├── 9-dimensions.md                Intent extraction framework\n    └── brief-formats.md               Ready-to-use templates per agent type\n```\n\n---\n\n## License\n\nMIT - see [LICENSE](LICENSE)\n\nFile v0.2.0:_meta.json\n\n{\n  \"ownerId\": \"kn7d3r58cdxk8k0xg6jq6k65gs873c0c\",\n  \"slug\": \"brief-master\",\n  \"version\": \"0.2.0\",\n  \"publishedAt\": 1779469505301\n}\n\nFile v0.2.0:references/9-dimensions.md\n\n# 9 Dimensions of Intent\n\nExtract only the dimensions needed to make the brief executable. Do not turn this into an interview script when context is already clear.\n\n## 1. Task\n\nWhat should the agent do?\n\n- One verb, one outcome\n- Split unrelated tasks into separate briefs\n- Prefer \"fix X\" / \"implement Y\" / \"review Z\" over vague \"work on\"\n\n## 2. Input\n\nWhat does the agent need to start?\n\n- Repo, files, URLs, datasets, credentials, or screenshots\n- Runtime/host if relevant\n- Branch, directory, service, or environment\n\nAvoid vague location text like \"in the repo\" when the exact path or branch matters.\n\n## 3. Output\n\nWhat should exist when the agent is done?\n\n- Files created or changed\n- Tests/proof artifacts generated\n- Commit, PR, report, or message expected\n- Service or data state expected\n\n## 4. Constraints\n\nWhat must not break?\n\n- Existing behaviours\n- Files, data, credentials, or generated assets that must not be touched\n- Backwards compatibility or performance limits\n- Security/privacy boundaries\n\n## 5. Context\n\nWhat does the agent need to understand about why this task exists?\n\nKeep it short. Include only context that changes the work.\n\n## 6. Audience / Role\n\nWho is the brief for?\n\n- Builder / implementer\n- Reviewer / verifier\n- Researcher\n- Cron or scheduled agent\n- Human operator\n\nRole determines autonomy, tool expectations, and output shape.\n\n## 7. Prior Context\n\nWhat should the agent read first?\n\n- Specs, issue links, previous verdicts, relevant docs\n- Existing implementation patterns\n- Prior decisions or non-goals\n\n## 8. Success Criteria\n\nHow does everyone know it is done?\n\n- Use explicit AC1, AC2, AC3 where possible\n- Each AC must be testable by a fresh verifier\n- Include the exact verification command or manual check when known\n\nIf you cannot write a testable AC, the task is under-defined.\n\n## 9. Examples\n\nWhat examples clarify the intended result?\n\n- Before/after screenshots\n- Expected output format\n- Similar accepted implementation\n- Known-good command output\n\n## Quick Extraction Prompt\n\nAsk at most three questions in one round:\n\n```text\nI can write the brief, but I need three specifics:\n1. [missing blocker]\n2. [missing blocker]\n3. [missing blocker]\n```\n\nIf more than three answers are missing, ask the three that most affect correctness and note the remaining assumptions in the brief.\n\nFile v0.2.0:references/brief-formats.md\n\n# Brief Formats\n\nUse these as public-safe templates. Replace placeholders with the user's actual environment only when the user has provided it or it is already part of the current task context.\n\n## Builder Brief\n\n```markdown\n# Brief: [TASK_ID] - [one-line outcome]\n\n## Confidence Gate\n\nBefore starting implementation, ask clarifying questions until you are 95% confident you can complete this task correctly. If you are already 95% confident, state assumptions and proceed.\n\n## Task\n\n[One clear sentence describing what to build or fix.]\n\n## Context\n\n[Only context that changes the implementation. 2-5 sentences.]\n\n## Acceptance Criteria\n\nAC1: [specific, testable condition]\nVerify: [command/check]\n\nAC2: [specific, testable condition]\nVerify: [command/check]\n\nAC3: [specific, testable condition]\nVerify: [command/check]\n\n## Constraints\n\n- [What must not break]\n- [Files/data/services that must not be touched]\n- [Security/privacy boundaries]\n\n## Non-Goals\n\n- [Explicitly out of scope]\n\n## Read First\n\n- [file/doc/spec/issue link]\n\n## Environment\n\n- Repo/path: [repo/path]\n- Branch: [branch]\n- Runtime/service: [if relevant]\n\n## Deliver\n\n- Changed files\n- Verification output\n- Known residual risks\n```\n\n## Reviewer / Verifier Brief\n\n```markdown\n# Review: [TASK_ID]\n\n## Role\n\nYou are the verifier. Do not implement fixes. Verify the acceptance criteria from a fresh session and report evidence.\n\n## Read First\n\n- [spec or brief]\n- [existing verdict/problems if any]\n\n## Acceptance Criteria To Verify\n\nAC1: [copy from spec]\nAC2: [copy from spec]\nAC3: [copy from spec]\n\n## Required Checks\n\n- [test command]\n- [manual/browser/API check]\n- [regression check]\n\n## Output\n\nReturn:\n- PASS/FAIL/UNKNOWN for each AC\n- Evidence for each verdict\n- File/line references for failures when applicable\n- Test commands run and exact result\n```\n\n## Research Brief\n\n```markdown\n# Research: [topic]\n\n## Question\n\n[Specific question to answer.]\n\n## Scope\n\n- Include: [source types, date range, domains, handles]\n- Exclude: [known noise]\n\n## Source Quality\n\n- Prefer primary sources.\n- Include URLs and publication dates.\n- Separate observed facts from inference.\n- Flag uncertainty.\n\n## Output\n\n- Short answer\n- Evidence table\n- Implications\n- Open questions\n```\n\n## Cron / Scheduled Agent Brief\n\n```markdown\n# Scheduled Task: [name]\n\n## Goal\n\n[What should be checked or produced.]\n\n## Steps\n\n1. [Exact action]\n2. [Exact action]\n3. [Exact action]\n\n## Success Condition\n\n[What normal healthy output means.]\n\n## Alert Condition\n\nSend a message only if:\n- [condition]\n- [condition]\n\nIf everything is healthy, return `NO_REPLY`.\n\n## Output\n\n- Write report to [path] if needed.\n- Include exact failure evidence if alerting.\n```\n\n## Token Efficiency Checklist\n\n- Can any sentence be removed without changing what the agent does?\n- Are paths, branches, commands, and URLs specific where needed?\n- Are ACs testable by someone who did not write the code?\n- Are constraints and non-goals explicit?\n- Are private local assumptions excluded unless required?\n- Is the final output shape clear?\n\nFile v0.2.0:LICENSE\n\nMIT License\n\nCopyright (c) 2026 Leo Stehlik\n\nPermission is hereby granted, free of charge, to any person obtaining a copy\nof this software and associated documentation files (the \"Software\"), to deal\nin the Software without restriction, including without limitation the rights\nto use, copy, modify, merge, publish, distribute, sublicense, and/or sell\ncopies of the Software, and to permit persons to whom the Software is\nfurnished to do so, subject to the following conditions:\n\nThe above copyright notice and this permission notice shall be included in all\ncopies or substantial portions of the Software.\n\nTHE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\nIMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\nFITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE\nAUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\nLIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,\nOUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE\nSOFTWARE.\n\nArchive v0.1.0: 6 files, 7988 bytes\n\nFiles: LICENSE (1068b), README.md (2667b), references/9-dimensions.md (3274b), references/brief-formats.md (4717b), SKILL.md (2800b), _meta.json (131b)\n\nFile v0.1.0:SKILL.md\n\n---\nname: brief-master\ndescription: Writes sharp, precise agent briefs and prompts for OpenClaw agents and cron jobs. Use when asked to write a brief for any agent — dev agents, code reviewers, testers, researchers — or when writing a cron job prompt, a sessions_spawn task, or any instruction that will be executed by an AI agent. Extracts 9 dimensions of intent, asks max 3 clarifying questions, runs a token efficiency audit, and delivers one clean brief. Prevents vague briefs that cause agent failures. Triggers on \"write a brief\", \"write a prompt for\", \"help me write the agent brief\", \"draft the cron job\", \"write the sessions_spawn task\".\n---\n\n# Brief Master\n\nWrite agent briefs that agents actually execute correctly.\n\nEvery wasted token is a wasted API call. Every vague word is a future bug. Every missing acceptance criterion is a future rework cycle.\n\nRead `references/9-dimensions.md` before extracting intent.\nRead `references/brief-formats.md` for the right format per agent type.\n\n## The Pipeline\n\n1. **Detect the target** — which agent, what runtime (subagent, cron, sessions_spawn)?\n2. **Extract 9 dimensions** — see `references/9-dimensions.md`\n3. **Ask max 3 questions** — only if critical info is missing. Never more.\n4. **Apply the right format** — see `references/brief-formats.md`\n5. **Run token efficiency audit** — strip every word that doesn't change the output\n6. **Deliver** — one clean brief, ready to use\n\n## Token Efficiency Rule\n\n\"The best brief is not the longest. It's the one where every word is load-bearing.\"\n\nBefore delivering, ask: does removing this sentence change what the agent does? If not, cut it.\n\n## What Makes a Bad Brief\n\n- Vague task description (\"improve the translation\")\n- No acceptance criteria (how does the agent know it's done?)\n- Missing constraints (what must not break?)\n- No non-goals (what's explicitly out of scope?)\n- Too long (agent loses focus, context drifts)\n- Tool-specific instructions missing (which host? which directory? which branch?)\n\n## What Makes a Good Brief\n\n- One clear task per brief\n- Explicit ACs labelled AC1, AC2, AC3 (testable, not descriptions)\n- Constraints listed\n- Non-goals listed\n- Correct status update blocks (start + end)\n- 95% confidence gate instruction included\n- Right tool commands with correct paths\n\n## Mandatory Sections for Dev Agent Briefs\n\nEvery brief for a dev/build agent must include:\n\n```\nFIRST — update status to working with task description\n...task...\nLAST — update status to done + notify orchestrator\n```\n\nAnd the 95% confidence gate:\n```\nBefore starting ANY work, ask clarifying questions until you are 95% confident\nyou can complete this task successfully. Do not start until you have that confidence.\n```\n\nSee `references/brief-formats.md` for full templates.\n\nFile v0.1.0:README.md\n\n# brief-master\n\nWrite sharp, precise agent briefs for OpenClaw. Zero wasted tokens. Zero vague instructions.\n\nEvery bad brief costs you a rework cycle. brief-master extracts 9 dimensions of intent, asks max 3 clarifying questions, applies the right format for your target agent, runs a token efficiency audit, and delivers one clean brief ready to fire.\n\nBuilt for OpenClaw's multi-agent workflow: dev agents, code reviewers, testers, researchers, and cron jobs.\n\nInspired by [prompt-master](https://github.com/nidhinjs/prompt-master). Built as our own OpenClaw-native implementation.\n\n---\n\n## The Problem\n\nEvery wasted token is a wasted API call.\nEvery vague word is a future bug.\nEvery missing acceptance criterion is a future rework cycle.\n\n\"Improve the translation system\" is not a brief.\n\"Fix formSchemaTranslated being NULL for German locale — AC1: all 50 prompt rows have formSchemaTranslated populated after running translate/de endpoint\" is a brief.\n\n---\n\n## The Pipeline\n\n1. Detect the target agent and runtime\n2. Extract 9 dimensions of intent\n3. Ask max 3 clarifying questions (only if critical info is missing)\n4. Apply the right format for the target agent\n5. Run token efficiency audit\n6. Deliver one clean, ready-to-use brief\n\n---\n\n## The 9 Dimensions\n\n1. **Task** - what exactly should be done (one verb, one outcome)\n2. **Input** - what does the agent need to start (files, machines, branches)\n3. **Output** - what should exist when done\n4. **Constraints** - what must not break\n5. **Context** - only what changes what the agent does\n6. **Audience/Role** - which agent, what runtime\n7. **Memory** - what files to read first\n8. **Success Criteria** - AC1, AC2, AC3 (testable, specific)\n9. **Examples** - reference points that clarify expected behaviour\n\n---\n\n## Installation\n\n### OpenClaw\n\n```json\n{\n  \"skills\": {\n    \"load\": {\n      \"extraDirs\": [\"/path/to/your/skills\"]\n    }\n  }\n}\n```\n\n```bash\ngit clone https://github.com/LeoStehlik/brief-master.git /path/to/your/skills/brief-master\n```\n\n### Claude Code / Codex\n\nCopy into `.agents/skills/brief-master/` or `.claude/skills/brief-master/`.\n\n---\n\n## Usage\n\n```\nWrite me a brief for my dev agent to fix the translation timeout issue\n```\n\n```\nHelp me write a cron job prompt for overnight research\n```\n\n```\nI need a code review brief for Sprint 5\n```\n\n---\n\n## What's Inside\n\n```\nbrief-master/\n├── SKILL.md                           Core pipeline + quality rules\n└── references/\n    ├── 9-dimensions.md                Intent extraction framework\n    └── brief-formats.md               Ready-to-use templates per agent type\n```\n\n---\n\n## License\n\nMIT - see [LICENSE](LICENSE)\n\nFile v0.1.0:_meta.json\n\n{\n  \"ownerId\": \"kn7d3r58cdxk8k0xg6jq6k65gs873c0c\",\n  \"slug\": \"brief-master\",\n  \"version\": \"0.1.0\",\n  \"publishedAt\": 1779320516714\n}\n\nFile v0.1.0:references/9-dimensions.md\n\n# 9 Dimensions of Intent\n\nExtract these before writing any brief. Every dimension that's missing is a potential failure point.\n\n---\n\n## 1. Task\n**What is the agent being asked to DO?**\n- One verb, one outcome\n- \"Implement X\", \"Fix Y\", \"Research Z\", \"Review W\"\n- If there are multiple tasks: split into multiple briefs\n\n**Bad:** \"Work on the translation system\"\n**Good:** \"Fix the formSchemaTranslated field not being populated for the target locale\"\n\n---\n\n## 2. Input\n**What does the agent need to start?**\n- Files, repos, URLs, data, credentials\n- Which machine or runtime? (local checkout, remote host, CI runner, browser session)\n- Which branch, directory, service?\n\n**Example:** `cd /path/to/repo && git status --short` — not just \"in the repo\"\n\n---\n\n## 3. Output\n**What should exist when the agent is done?**\n- Files created/modified\n- Services restarted?\n- DB records updated?\n- Git commit pushed?\n- Status updated?\n\n---\n\n## 4. Constraints\n**What must NOT break?**\n- Existing functionality (specify which features/sprints)\n- Files that must not be overwritten\n- DB records that must not be deleted\n- Services that must stay running\n\n---\n\n## 5. Context\n**What does the agent need to understand about the broader situation?**\n- Why this task exists\n- What failed before\n- What was tried already\n- Architecture decisions that affect this task\n\nKeep this SHORT. Only context that changes what the agent does.\n\n---\n\n## 6. Audience / Role\n**Who is this brief for?**\n- Builder agent (implementation)\n- Reviewer/verifier agent (code review, tests, live browser checks — NOT implementation)\n- Fixer agent (minimal follow-up fixes from verifier findings)\n- Research agent (source-backed investigation)\n- Cron/scheduled job (automated recurring task)\n- Human operator (final decision or risky external action)\n\nRole determines: tone, expected tools, level of autonomy, what they should ask vs decide.\n\n---\n\n## 7. Memory / Prior Context\n**What should the agent read before starting?**\n- Relevant context files\n- Prior commits or decisions\n- The spec.md if this is a proof-loop task\n- Code dependency/blast radius analysis\n\n---\n\n## 8. Success Criteria\n**How does the agent (and you) know it's done?**\n- AC1, AC2, AC3 — explicit, testable, specific\n- Regression suite must pass?\n- Visual verification required?\n- DB state expected?\n\nIf you can't write a testable AC, the task isn't defined well enough yet.\n\n---\n\n## 9. Examples\n**Are there examples that clarify the expected behaviour?**\n- Before/after screenshots\n- Expected output format\n- Similar completed tasks to reference\n- Known-good code patterns to follow\n\n---\n\n## Quick Extraction Prompt\n\nWhen someone asks for help writing a brief, ask:\n\n> \"Before I write this brief, let me extract the key dimensions:\n> 1. What exactly should be done? (one sentence)\n> 2. What does the agent need access to? (files, machines, branches)\n> 3. What should exist when it's done?\n> 4. What must not break?\n> 5. Any critical context the agent needs?\n> 6. Which agent/role is this for?\n> 7. Which files/docs should the agent read first?\n> 8. What are the acceptance criteria? (AC1, AC2, AC3)\n> 9. Any examples or reference points?\"\n\nAsk only the questions where the answer isn't already obvious from context.\nMax 3 questions per round.\n\nFile v0.1.0:references/brief-formats.md\n\n# Brief Formats by Agent Type\n\nUse these templates as public-safe starting points. Replace placeholders before sending a brief to an agent.\n\n---\n\n## Campaign Setup\n\nUse this when a task needs multiple agents or multiple phases.\n\n```markdown\n# Campaign: [CAMPAIGN_ID]\n\n## Goal\n[One paragraph. What does done look like?]\n\n## Phases\n1. Spec freeze: [owner]\n2. Build: [owner]\n3. Verify: [fresh verifier owner]\n4. Fix loop: [owner, if needed]\n\n## Current State\n- Active phase:\n- Repo/path:\n- Branch:\n- Last update:\n\n## Discovery Log\n- [Timestamp] [Fact future agents must know]\n```\n\nEvery agent brief in the campaign should say:\n\n```markdown\nRead the campaign file first.\nUpdate the discovery log when you learn something important that future agents must know.\n```\n\n---\n\n## Builder Brief\n\n```markdown\n# Brief: [TASK_ID] — [One-line task]\n\nBefore starting ANY work, ask clarifying questions until you are 95% confident\nyou can complete this task successfully. Do not start until you have that confidence.\n\n## Task\n[One clear sentence describing what needs to be built/fixed]\n\n## Context\n[Only what changes what the agent does — 2-5 sentences max]\n\n## Acceptance Criteria\nAC1: [specific, testable condition]\n     Verify: [how to check]\nAC2: [specific, testable condition]\n     Verify: [how to check]\nAC3: [specific, testable condition]\n     Verify: [how to check]\n\n## Constraints\n- [What must not break]\n- [Files/services that must not be touched]\n\n## Non-Goals\n- [Explicitly out of scope]\n\n## Read First\n- [Relevant spec, issue, docs, or files]\n\n## Environment\n- Repo/path: [repo URL or local path]\n- Branch: [branch name]\n- Commands: [test/build/lint commands]\n\n## Output\n- List changed files\n- Summarize implementation\n- Include exact verification commands and results\n- Do not claim done unless every AC is verified or explicitly marked blocked\n```\n\n---\n\n## Fresh Verifier Brief\n\n```markdown\n# Verify: [TASK_ID]\n\nYou are the verifier. You did not build this work.\nDo not edit production code. Do not sign off based only on the builder's summary.\n\n## Read First\n- [Frozen spec or task brief]\n- [Builder evidence, if any]\n- [Known problems, if any]\n\n## Acceptance Criteria to Verify\nAC1: [copy exactly]\nAC2: [copy exactly]\nAC3: [copy exactly]\n\n## Required Checks\n- Run: [test command]\n- Inspect: [files or behavior]\n- If UI work: verify in browser and capture a screenshot or clear observation\n\n## Verdict Output\nWrite or return:\n\n```json\n{\n  \"task_id\": \"[TASK_ID]\",\n  \"phase\": \"verify\",\n  \"agent\": \"verifier\",\n  \"timestamp\": \"[ISO-8601]\",\n  \"overall\": \"PASS|FAIL|UNKNOWN\",\n  \"criteria\": [\n    { \"id\": \"AC1\", \"status\": \"PASS|FAIL|UNKNOWN\", \"note\": \"[evidence]\" }\n  ],\n  \"problems\": [\n    { \"criterion\": \"AC2\", \"problem\": \"[specific issue]\", \"evidence\": \"[file/line, command output, or browser observation]\" }\n  ]\n}\n```\n```\n\n---\n\n## Fixer Brief\n\n```markdown\n# Fix: [TASK_ID]\n\nRead the verifier output first.\nFix only the failures the verifier identified.\nDo not broaden scope.\n\n## Known Problems\n- [AC/id]: [specific problem]\n\n## Constraints\n- Preserve passing ACs\n- Keep the diff minimal\n- Re-run the verification commands affected by the fix\n\n## Output\n- Changed files\n- Problem-to-fix mapping\n- Verification results\n- Any remaining blocked/unknown items\n```\n\n---\n\n## Research Brief\n\n```markdown\n# Research: [TOPIC]\n\n## Question\n[Specific question to answer]\n\n## Scope\n- Include:\n- Exclude:\n- Time horizon:\n\n## Source Quality\n- Prefer primary sources, official docs, papers, repos, changelogs, or first-hand posts\n- Include URL and publication/update date for every source\n- Separate observed facts from inference\n\n## Output\n- Executive answer in 3-5 bullets\n- Evidence table with sources\n- Risks/unknowns\n- Recommended next action\n```\n\n---\n\n## Scheduled Job / Cron Brief\n\n```markdown\n# Scheduled Job: [JOB_NAME]\n\n## Purpose\n[What this recurring task is supposed to protect or produce]\n\n## Schedule\n[Frequency and timezone]\n\n## Steps\n1. [Exact command/API/tool action]\n2. [Validation check]\n3. [Write output/report path, if any]\n\n## Normal Result\nIf everything is expected: [NO_REPLY / write report / send summary]\n\n## Alert Conditions\nSend an alert only when:\n- [Condition]\n- [Condition]\n\n## Recovery\n- Retry policy:\n- Escalation condition:\n- Kill switch:\n```\n\n---\n\n## Token Efficiency Checklist\n\nBefore delivering any brief, check:\n- [ ] Can any sentence be removed without changing what the agent does?\n- [ ] Are all paths explicit enough for the target runtime?\n- [ ] Are machine/runtime references explicit?\n- [ ] Are ACs testable by someone who did not write the code?\n- [ ] Is the 95% confidence gate included when useful?\n- [ ] Are non-goals and constraints included?\n- [ ] Is the output format unambiguous?\n\nFile v0.1.0:LICENSE\n\nMIT License\n\nCopyright (c) 2026 Leo Stehlik\n\nPermission is hereby granted, free of charge, to any person obtaining a copy\nof this software and associated documentation files (the \"Software\"), to deal\nin the Software without restriction, including without limitation the rights\nto use, copy, modify, merge, publish, distribute, sublicense, and/or sell\ncopies of the Software, and to permit persons to whom the Software is\nfurnished to do so, subject to the following conditions:\n\nThe above copyright notice and this permission notice shall be included in all\ncopies or substantial portions of the Software.\n\nTHE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\nIMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\nFITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE\nAUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\nLIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,\nOUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE\nSOFTWARE.","readmeExcerpt":"Skill: Brief Master Owner: leostehlik Summary: Write precise execution briefs for agents, cron jobs, reviewers, researchers, and delegated coding tasks. Tags: agents:0.2.1, briefs:0.2.1, latest:0.2.1, openclaw:0.2.1, prompting:0.2.1 Version history: v0.2.1 | 2026-05-24T10:00:02.661Z | user Document that Brief Master drafts briefs only, requires review before autonomous execution, and must not invent credentials, appr","codeSnippets":[],"executableExamples":[{"language":"text","snippet":"FIRST — update status to working with task description\n...task...\nLAST — update status to done + notify orchestrator"},{"language":"text","snippet":"Before starting ANY work, ask clarifying questions until you are 95% confident\nyou can complete this task successfully. Do not start until you have that confidence."},{"language":"json","snippet":"{\n  \"skills\": {\n    \"load\": {\n      \"extraDirs\": [\"/path/to/your/skills\"]\n    }\n  }\n}"},{"language":"bash","snippet":"git clone https://github.com/LeoStehlik/brief-master.git /path/to/your/skills/brief-master"},{"language":"text","snippet":"Write me a brief for my dev agent to fix the translation timeout issue"},{"language":"text","snippet":"Help me write a cron job prompt for overnight research"}],"parameters":null,"dependencies":[],"permissions":[],"extractedFiles":[{"path":"SKILL.md","content":"---\nname: brief-master\ndescription: \"Write precise execution briefs for agents, cron jobs, reviewers, researchers, and delegated coding tasks.\"\nmetadata:\n  version: \"0.2.1\"\n---\n# Brief Master\n\nWrite agent briefs that agents actually execute correctly.\n\nEvery wasted token is a wasted API call. Every vague word is a future bug. Every missing acceptance criterion is a future rework cycle.\n\nRead `references/9-dimensions.md` before extracting intent.\nRead `references/brief-formats.md` for the right format per agent type.\n\n\n## Safety Boundary\n\nThis skill writes briefs only. A generated brief can instruct another agent to edit code, run commands, schedule jobs, or use external services, so the user or orchestrator must review the final brief before execution.\n\nDo not invent credentials, secret names, hostnames, destructive commands, cron schedules, repository permissions, or approval status. If a brief would require elevated access, remote execution, persistent automation, or public posting, mark that as requiring explicit approval in the brief.\n\n## The Pipeline\n\n1. **Detect the target** — which agent, what runtime (subagent, cron, sessions_spawn)?\n2. **Extract 9 dimensions** — see `references/9-dimensions.md`\n3. **Ask max 3 questions** — only if critical info is missing. Never more.\n4. **Apply the right format** — see `references/brief-formats.md`\n5. **Run token efficiency audit** — strip every word that doesn't change the output\n6. **Deliver** — one clean brief, ready to use\n\n## Token Efficiency Rule\n\n\"The best brief is not the longest. It's the one where every word is load-bearing.\"\n\nBefore delivering, ask: does removing this sentence change what the agent does? If not, cut it.\n\n## What Makes a Bad Brief\n\n- Vague task description (\"improve the translation\")\n- No acceptance criteria (how does the agent know it's done?)\n- Missing constraints (what must not break?)\n- No non-goals (what's explicitly out of scope?)\n- Too long (agent loses focus, context drifts)\n- Tool-specific instructions missing (which host? which directory? which branch?)\n\n## What Makes a Good Brief\n\n- One clear task per brief\n- Explicit ACs labelled AC1, AC2, AC3 (testable, not descriptions)\n- Constraints listed\n- Non-goals listed\n- Correct status update blocks (start + end)\n- 95% confidence gate instruction included\n- Right tool commands with correct paths\n\n## Mandatory Sections for Dev Agent Briefs\n\nEvery brief for a dev/build agent must include:\n\n```\nFIRST — update status to working with task description\n...task...\nLAST — update status to done + notify orchestrator\n```\n\nAnd the 95% confidence gate:\n```\nBefore starting ANY work, ask clarifying questions until you are 95% confident\nyou can complete this task successfully. Do not start until you have that confidence.\n```\n\nSee `references/brief-formats.md` for full templates."},{"path":"examples/README.md","content":"# Brief Master Examples\n\nStart with [`messy-to-clean-brief.md`](messy-to-clean-brief.md).\n\nIt shows the core move of Brief Master: take a fuzzy request, preserve the intent, and turn it into an agent brief with scope, constraints, acceptance criteria, non-goals, and verification commands.\n\nUse it when you want to see what changes between \"please improve this\" and a brief an implementation agent can actually execute."},{"path":"README.md","content":"# brief-master\n\n![Validate](https://github.com/LeoStehlik/brief-master/actions/workflows/validate.yml/badge.svg)\n\nWrite sharp, precise agent briefs for OpenClaw. Zero wasted tokens. Zero vague instructions.\n\nBrief Master is an agent brief writer for AI coding workflows. It extracts 9 dimensions of intent, asks max 3 clarifying questions, applies the right format for your target agent, runs a token efficiency audit, and delivers one clean brief ready to fire.\n\nBuilt for OpenClaw's multi-agent workflow: dev agents, code reviewers, testers, researchers, and cron jobs.\n\nInspired by [prompt-master](https://github.com/nidhinjs/prompt-master). Built as our own OpenClaw-native implementation.\n\n\n## Safety Boundary\n\nBrief Master only drafts execution briefs. Review the generated brief before handing it to an autonomous agent, especially commands, file paths, credentials, scheduled jobs, external services, and anything that changes code or system state.\n\nThe skill must not invent access, approval, hostnames, secrets, destructive commands, or cron schedules. If execution needs elevated privileges or persistent automation, the brief should say that explicit approval is required.\n\n## Use Cases\n\n- turn a messy request into a precise agent brief\n- make acceptance criteria explicit before an agent starts work\n- capture constraints, non-goals, inputs, and verification steps\n- reduce rework from vague prompts and missing context\n\n---\n\n## The Problem\n\nEvery wasted token is a wasted API call.\nEvery vague word is a future bug.\nEvery missing acceptance criterion is a future rework cycle.\n\n\"Improve the translation system\" is not a brief.\n\"Fix formSchemaTranslated being NULL for German locale — AC1: all 50 prompt rows have formSchemaTranslated populated after running translate/de endpoint\" is a brief.\n\n---\n\n## Credibility Artifact\n\nSee [`examples/README.md`](examples/README.md) for the example index, or jump straight to [`examples/messy-to-clean-brief.md`](examples/messy-to-clean-brief.md) for a concrete messy request -> clean agent brief example with ACs, constraints, non-goals, and verification commands.\n\n## The Pipeline\n\n1. Detect the target agent and runtime\n2. Extract 9 dimensions of intent\n3. Ask max 3 clarifying questions (only if critical info is missing)\n4. Apply the right format for the target agent\n5. Run token efficiency audit\n6. Deliver one clean, ready-to-use brief\n\n---\n\n## The 9 Dimensions\n\n1. **Task** - what exactly should be done (one verb, one outcome)\n2. **Input** - what does the agent need to start (files, machines, branches)\n3. **Output** - what should exist when done\n4. **Constraints** - what must not break\n5. **Context** - only what changes what the agent does\n6. **Audience/Role** - which agent, what runtime\n7. **Memory** - what files to read first\n8. **Success Criteria** - AC1, AC2, AC3 (testable, specific)\n9. **Examples** - reference points that clarify expected behaviour\n\n---\n\n\n## When To Use Which Repo\n\nUse this repo when the task is still too "},{"path":"_meta.json","content":"{\n  \"ownerId\": \"kn7d3r58cdxk8k0xg6jq6k65gs873c0c\",\n  \"slug\": \"brief-master\",\n  \"version\": \"0.2.1\",\n  \"publishedAt\": 1779616802661\n}"},{"path":"references/9-dimensions.md","content":"# 9 Dimensions of Intent\n\nExtract only the dimensions needed to make the brief executable. Do not turn this into an interview script when context is already clear.\n\n## 1. Task\n\nWhat should the agent do?\n\n- One verb, one outcome\n- Split unrelated tasks into separate briefs\n- Prefer \"fix X\" / \"implement Y\" / \"review Z\" over vague \"work on\"\n\n## 2. Input\n\nWhat does the agent need to start?\n\n- Repo, files, URLs, datasets, credentials, or screenshots\n- Runtime/host if relevant\n- Branch, directory, service, or environment\n\nAvoid vague location text like \"in the repo\" when the exact path or branch matters.\n\n## 3. Output\n\nWhat should exist when the agent is done?\n\n- Files created or changed\n- Tests/proof artifacts generated\n- Commit, PR, report, or message expected\n- Service or data state expected\n\n## 4. Constraints\n\nWhat must not break?\n\n- Existing behaviours\n- Files, data, credentials, or generated assets that must not be touched\n- Backwards compatibility or performance limits\n- Security/privacy boundaries\n\n## 5. Context\n\nWhat does the agent need to understand about why this task exists?\n\nKeep it short. Include only context that changes the work.\n\n## 6. Audience / Role\n\nWho is the brief for?\n\n- Builder / implementer\n- Reviewer / verifier\n- Researcher\n- Cron or scheduled agent\n- Human operator\n\nRole determines autonomy, tool expectations, and output shape.\n\n## 7. Prior Context\n\nWhat should the agent read first?\n\n- Specs, issue links, previous verdicts, relevant docs\n- Existing implementation patterns\n- Prior decisions or non-goals\n\n## 8. Success Criteria\n\nHow does everyone know it is done?\n\n- Use explicit AC1, AC2, AC3 where possible\n- Each AC must be testable by a fresh verifier\n- Include the exact verification command or manual check when known\n\nIf you cannot write a testable AC, the task is under-defined.\n\n## 9. Examples\n\nWhat examples clarify the intended result?\n\n- Before/after screenshots\n- Expected output format\n- Similar accepted implementation\n- Known-good command output\n\n## Quick Extraction Prompt\n\nAsk at most three questions in one round:\n\n```text\nI can write the brief, but I need three specifics:\n1. [missing blocker]\n2. [missing blocker]\n3. [missing blocker]\n```\n\nIf more than three answers are missing, ask the three that most affect correctness and note the remaining assumptions in the brief."}],"languages":[],"docsSourceLabel":"CLAWHUB","editorialOverview":null,"editorialQuality":{"score":100,"threshold":65,"status":"thin","wordCount":1550,"uniquenessScore":44,"reasons":["uniqueness-below-45"]}},"media":{"evidence":{"source":"no-media","verified":false,"confidence":"low","updatedAt":"2026-10-11T22:47:00.838Z","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":"2026-10-11T22:47:00.838Z","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":"2026-10-12T16:27:00.989Z","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"}]}}}