{"id":"4c647bc5-9281-47e0-b267-2d48a2c5ed32","entityType":"agent","slug":"clawhub-hhhwor-chukonu-web-search","name":"chukonu-web-search","canonicalUrl":"https://www.xpersona.co/agent/clawhub-hhhwor-chukonu-web-search","canonicalPath":"/agent/clawhub-hhhwor-chukonu-web-search","generatedAt":"2026-10-11T14:15:38.342Z","source":"CLAWHUB","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"editorial-content","verified":true,"confidence":"high","updatedAt":"2026-10-11T11:29:35.766Z","emptyReason":null},"description":"通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利、金融和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及财经新闻与金融研究、法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。 Skill: chukonu-web-search Owner: hhhwor Summary: 通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利、金融和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及财经新闻与金融研究、法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。 Tags: latest:1.1.4 Version history: v1.1.4 | 2026-09-09T01:02:26.993Z | user chukonu-web-search 1.1.4 - 新增财经新闻与金融文档检索及研究支持，source_types 增加 financial，支持金融新闻、结构化行情、财报等能力。 - 文档指引","descriptionLabel":"Technical summary","evidenceSummary":"Capability contract not published. No trust telemetry is available yet. 1.1K downloads reported by the source. Last updated 10/11/2026.","installCommand":"clawhub skill install s17cqswjp1a794wqcd41r5x4px8a29kx:chukonu-web-search","sourceUrl":"https://clawhub.ai/hhhwor/chukonu-web-search","homepage":"https://clawhub.ai/hhhwor/skills/chukonu-web-search","primaryLinks":[{"label":"View on ClawHub","url":"https://clawhub.ai/hhhwor/chukonu-web-search","kind":"source"},{"label":"Homepage","url":"https://clawhub.ai/hhhwor/skills/chukonu-web-search","kind":"homepage"}],"safetyScore":84,"overallRank":62,"popularityScore":61,"trustScore":null,"claimedByName":null,"isOwner":false,"seoDescription":"通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利、金融和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及财经新闻与金融研究、法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieva"},"coverage":{"evidence":{"source":"public-profile","verified":false,"confidence":"medium","updatedAt":"2026-10-11T11:29:35.766Z","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":"2026-10-11T11:29:35.766Z","emptyReason":null},"stars":null,"forks":null,"downloads":1076,"packageName":null,"latestVersion":"1.1.4","tractionLabel":"1.1K downloads"},"release":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T11:29:35.692Z","emptyReason":null},"lastUpdatedAt":"2026-10-11T11:29:35.766Z","lastCrawledAt":"2026-10-11T11:29:35.692Z","lastIndexedAt":null,"nextCrawlAt":"2026-10-12T11:29:35.692Z","lastVerifiedAt":null,"highlights":[{"version":"1.1.4","createdAt":"2026-09-09T01:02:26.993Z","changelog":"chukonu-web-search 1.1.4 - 新增财经新闻与金融文档检索及研究支持，`source_types` 增加 `financial`，支持金融新闻、结构化行情、财报等能力。 - 文档指引细化金融搜索、财经新闻、金融研究操作流程和返回结果解读，包括意图、计划与数据来源区分。 - 零散梳理与强化法律法规、财经、专利、网页等多类来源查询和过滤使用说明，以及结果有效性、过滤器应用及限制项披露。 - 删除冗余 skill-card.md 文件，提升维护一致性。 - 更新接入和 OAuth 授权步骤，补充多平台认证配置注意事项。","fileCount":3,"zipByteSize":8730},{"version":"1.1.3","createdAt":"2026-08-11T01:47:27.497Z","changelog":"- Added support and documentation for retrieving Chinese legal and regulatory evidence via the internal FY MCP provider. - Updated usage guidelines and answer rules to cover legal evidence, including handling of legal source types, fields, and specific citation and limitation requirements. - Expanded tool selection instructions to include legal sources, and clarified handling for `source_types: legal`. - Added Codex CLI integration examples alongside OpenClaw and Claude Code. - Removed the `skill-card.md` file to simplify distribution.","fileCount":3,"zipByteSize":5832},{"version":"1.1.1","createdAt":"2026-08-04T00:46:57.935Z","changelog":"Version 1.1.1 - Updated version metadata from 0.3.0 to 0.3.1. - In the `search` result review process, added step to analyze `result_set.counts_by_stage` for diagnosing where candidate loss occurs. - Clarified not to directly connect to internal ports (e.g., 9001, 9101) or use internal API endpoints. - Removed the file: skill-card.md.","fileCount":3,"zipByteSize":6257},{"version":"1.1.0","createdAt":"2026-07-21T02:38:04.594Z","changelog":"**Chukonu Web Search 1.0.3 Changelog** - Updated and simplified skill setup instructions and documentation. - Clarified use of only public tools (`search`, `research`); removed references to internal-only endpoints. - Strengthened guidance on structured evidence handling, multi-phase research workflow, and research result interpretation. - Improved error, input, and boundary handling instructions for both search and research tasks. - Removed the redundant skill-card.md file for maintainability.","fileCount":3,"zipByteSize":6155},{"version":"1.0.2","createdAt":"2026-07-16T09:30:58.408Z","changelog":"- Removed the file: skill-card.md.","fileCount":3,"zipByteSize":5819},{"version":"1.0.1","createdAt":"2026-07-08T00:35:17.562Z","changelog":"**OAuth-only, multi-language update with improved setup and documentation** - Switched authentication to OAuth-only via MCP, removing static API tokens. - Added Chinese-language documentation and example workflows for OpenClaw and Claude Code. - Updated tool description, tags, and usage instructions for clarity and accuracy. - Clarified integration/setup steps and error handling for common authentication and usage issues. - No code changes; update is documentation- and configuration-focused.","fileCount":3,"zipByteSize":5015},{"version":"1.0.0","createdAt":"2026-07-07T20:31:33.356Z","changelog":"Initial release of Chukonu Web Search skill. - Enables structured web, academic, and patent search via https://search.houdutech.cn/web/. - Returns evidence-backed answers, highlighting answerability gaps and partial failures. - Connects through MCP or REST endpoints, with optional academic and patent filtering. - Covers only standard search; PDF full-text extraction not included. - Provides explicit guidance on evidence handling, citation, and answer moderation.","fileCount":3,"zipByteSize":3548}]},"execution":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No published capability contract is available yet."},"installCommand":"clawhub skill install s17cqswjp1a794wqcd41r5x4px8a29kx:chukonu-web-search","setupComplexity":"low","setupSteps":["Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.","Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data."],"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-hhhwor-chukonu-web-search/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-hhhwor-chukonu-web-search/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-hhhwor-chukonu-web-search/trust"},"curlExamples":["curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-hhhwor-chukonu-web-search/snapshot\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-hhhwor-chukonu-web-search/contract\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-hhhwor-chukonu-web-search/trust\""],"jsonRequestTemplate":{"query":"summarize this repo","constraints":{"maxLatencyMs":2000,"protocolPreference":["OPENCLEW"]}},"jsonResponseTemplate":{"ok":true,"result":{"summary":"...","confidence":0.9},"meta":{"source":"CLAWHUB","generatedAt":"2026-10-11T14:15:38.341Z"}},"retryPolicy":{"maxAttempts":3,"backoffMs":[500,1500,3500],"retryableConditions":["HTTP_429","HTTP_503","NETWORK_TIMEOUT"]}},"endpoints":{"dossierUrl":"https://www.xpersona.co/api/v1/agents/clawhub-hhhwor-chukonu-web-search/dossier","snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-hhhwor-chukonu-web-search/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-hhhwor-chukonu-web-search/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-hhhwor-chukonu-web-search/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":"high","updatedAt":"2026-10-11T11:29:35.766Z","emptyReason":null},"readme":"Skill: chukonu-web-search\n\nOwner: hhhwor\n\nSummary: 通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利、金融和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及财经新闻与金融研究、法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。\n\nTags: latest:1.1.4\n\nVersion history:\n\nv1.1.4 | 2026-09-09T01:02:26.993Z | user\n\nchukonu-web-search 1.1.4\n\n- 新增财经新闻与金融文档检索及研究支持，`source_types` 增加 `financial`，支持金融新闻、结构化行情、财报等能力。\n- 文档指引细化金融搜索、财经新闻、金融研究操作流程和返回结果解读，包括意图、计划与数据来源区分。\n- 零散梳理与强化法律法规、财经、专利、网页等多类来源查询和过滤使用说明，以及结果有效性、过滤器应用及限制项披露。\n- 删除冗余 skill-card.md 文件，提升维护一致性。\n- 更新接入和 OAuth 授权步骤，补充多平台认证配置注意事项。\n\nv1.1.3 | 2026-08-11T01:47:27.497Z | user\n\n- Added support and documentation for retrieving Chinese legal and regulatory evidence via the internal FY MCP provider.\n- Updated usage guidelines and answer rules to cover legal evidence, including handling of legal source types, fields, and specific citation and limitation requirements.\n- Expanded tool selection instructions to include legal sources, and clarified handling for `source_types: legal`.\n- Added Codex CLI integration examples alongside OpenClaw and Claude Code.\n- Removed the `skill-card.md` file to simplify distribution.\n\nv1.1.1 | 2026-08-04T00:46:57.935Z | user\n\nVersion 1.1.1\n\n- Updated version metadata from 0.3.0 to 0.3.1.\n- In the `search` result review process, added step to analyze `result_set.counts_by_stage` for diagnosing where candidate loss occurs.\n- Clarified not to directly connect to internal ports (e.g., 9001, 9101) or use internal API endpoints.\n- Removed the file: skill-card.md.\n\nv1.1.0 | 2026-07-21T02:38:04.594Z | user\n\n**Chukonu Web Search 1.0.3 Changelog**\n\n- Updated and simplified skill setup instructions and documentation.\n- Clarified use of only public tools (`search`, `research`); removed references to internal-only endpoints.\n- Strengthened guidance on structured evidence handling, multi-phase research workflow, and research result interpretation.\n- Improved error, input, and boundary handling instructions for both search and research tasks.\n- Removed the redundant skill-card.md file for maintainability.\n\nv1.0.2 | 2026-07-16T09:30:58.408Z | user\n\n- Removed the file: skill-card.md.\n\nv1.0.1 | 2026-07-08T00:35:17.562Z | user\n\n**OAuth-only, multi-language update with improved setup and documentation**\n\n- Switched authentication to OAuth-only via MCP, removing static API tokens.\n- Added Chinese-language documentation and example workflows for OpenClaw and Claude Code.\n- Updated tool description, tags, and usage instructions for clarity and accuracy.\n- Clarified integration/setup steps and error handling for common authentication and usage issues.\n- No code changes; update is documentation- and configuration-focused.\n\nv1.0.0 | 2026-07-07T20:31:33.356Z | auto\n\nInitial release of Chukonu Web Search skill.\n\n- Enables structured web, academic, and patent search via https://search.houdutech.cn/web/.\n- Returns evidence-backed answers, highlighting answerability gaps and partial failures.\n- Connects through MCP or REST endpoints, with optional academic and patent filtering.\n- Covers only standard search; PDF full-text extraction not included.\n- Provides explicit guidance on evidence handling, citation, and answer moderation.\n\nArchive index:\n\nArchive v1.1.4: 3 files, 8730 bytes\n\nFiles: skill-card.md (2475b), SKILL.md (15063b), _meta.json (137b)\n\nFile v1.1.4:SKILL.md\n\n---\nname: chukonu-web-search\ndescription: \"通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利、金融和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及财经新闻与金融研究、法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。\"\n---\n\n# Chukonu Search + Research\n\n通过 Chukonu remote MCP 调用公开的 `search` 和 `research` 工具。无需本地二进制、API key 或静态 token；不要调用内部工具、上游服务或内部端口。法律法规检索是服务内部接入的 FY MCP provider，客户端不要单独配置它、传入其凭据或尝试调用其工具。\n\n## 一次性接入\n\n使用 `https://search.houdutech.cn/web/mcp/` streamable-http 端点。该端点通过标准 OAuth 2.0 保护，采用动态客户端注册、授权码、PKCE(S256) 和 refresh token，scope 为 `search:read`。让 MCP host 自动发现并注册，不要手工填写 client ID 或 client secret。\n\nOpenClaw：\n\n```bash\nopenclaw mcp add chukonu-web-search \\\n  --url https://search.houdutech.cn/web/mcp/ \\\n  --transport streamable-http \\\n  --auth oauth\nopenclaw mcp login chukonu-web-search\n```\n\nClaude Code：\n\n```bash\nclaude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ \\\n  --scope user --transport http\n# 然后在会话中：/mcp → chukonu-web-search → Authenticate\n```\n\nCodex：\n\n```bash\ncodex mcp add chukonu-web-search --url https://search.houdutech.cn/web/mcp/\ncodex mcp login chukonu-web-search\ncodex mcp list\n```\n\n在 Codex TUI 中用 `/mcp` 查看连接状态。若 OAuth 浏览器与 Codex 不在同一主机，先为 Codex 配置可回调的 `mcp_oauth_callback_url`，再登录；不要输出或持久化授权码、回调 URL、access token 或 refresh token。\n\n不要为该服务器配置静态 `Authorization` 头或 `CHUKONU_SEARCH_API_TOKEN`。静态 Authorization 头会使 host 禁用 OAuth 回退，并可能导致 `401 invalid_token`。把 OAuth token 完全交给 host 保管，不要在日志、回答、示例或错误信息中泄露。\n\n## 选择工具与来源\n\n- 用 `search` 处理秒级发现、法规定位和获取后续研究所需的 `research_seed.search_id`。\n- 用 `research` 处理关键主张核验、反证、跨来源冲突、全文定位或证据覆盖评估。必须先 `search`，再用返回的 `search_id` 启动任务。\n- 本技能使用的 `source_types` 为 `web`、`academic`、`patent`、`legal` 和 `financial`。省略时服务自动路由；财经新闻归入 `financial`，法律意图可在法律 provider 启用时与通用网页并行召回。\n- 用 `source_types: [\"legal\"]` 只检索中国法律法规；用 `source_types: [\"legal\", \"web\"]` 同时获取法条和通用网页背景。\n- 用 `source_types: [\"financial\"]` 检索财经新闻和金融文档；宁夏服务已接入 `nws_news` 新闻库，其他金融来源取决于服务端配置。仅查金融时显式指定来源；需要网页原文或背景时用 `[\"financial\", \"web\"]`。\n- 金融问题需要进一步研究时，用 `research` 的 `profile: \"financial_research\"`；结构化股票数据是否可用要检查返回的 `financial_plan`，不能仅凭 profile 已接受就认定已启用行情查询。\n- 对法律问题尽量写明法名和条号，例如 `《中华人民共和国民法典》第1084条`；仅把原始法条证据用于法律效力或条文结论。\n- 不要把搜索相关性分数解释为事实置信度，也不要把检索结果或本技能当作法律意见。\n\n## `search`\n\n### 请求\n\n最小调用：\n\n```json\n{\n  \"query\": \"固态电池硫化物电解质近五年的关键路线\"\n}\n```\n\n只检索现行有效的中国法规：\n\n```json\n{\n  \"query\": \"《中华人民共和国民法典》第1084条\",\n  \"limit\": 10,\n  \"source_types\": [\"legal\"],\n  \"filters\": {\n    \"legal_status\": \"现行有效\",\n    \"jurisdictions\": [\"CN\"]\n  }\n}\n```\n\n混合检索法条和网页解读：\n\n```json\n{\n  \"query\": \"劳动合同解除的法定情形和近期解读\",\n  \"source_types\": [\"legal\", \"web\"],\n  \"filters\": {\n    \"legal_status\": \"现行有效\",\n    \"languages\": [\"zh\"]\n  }\n}\n```\n\n只检索财经新闻和金融文档：\n\n```json\n{\n  \"query\": \"贵州茅台 分红\",\n  \"source_types\": [\"financial\"],\n  \"filters\": {\n    \"published_from\": \"2026-08-01\",\n    \"published_to\": \"2026-08-31\"\n  }\n}\n```\n\n遵守以下边界：\n\n- `query` 必填；保持简洁、具体。`limit` 是最终全局返回数，范围为 1–20。\n- 本技能使用的 `filters` 包括 `published_from`、`published_to`、`languages`、`jurisdictions`、`legal_status` 和 `patent`。`legal_status` 仅接受 `尚未生效`、`现行有效`、`已被修改`、`失效`、`待核实`。\n- 日期、语言和法域过滤是否真正生效取决于来源。法律 provider 是中国法规来源，当前实际执行的是 `legal_status`；非中国法域和其他未支持过滤器会在响应中标明。不要仅因请求中带有过滤器就声称过滤已应用。\n- 不要发送请求级模型、重排、PDF 或 trust 开关；未知字段会被拒绝。\n\n### 金融垂直搜索\n\n- 查询中写明公司名称或已知证券代码，并加上具体主题，如分红、经营业绩、并购或行业政策；命中公司名称可能只是正文提及，需核对事件主体。\n- `published_from/published_to` 按新闻发布日期过滤，不代表财报报告期、交易日期或事件发生日期。示例日期须按用户问题调整；“最新”结论以实际 `published_date` 为准，空结果只说明本次检索未找到证据。\n- 当前宁夏新闻来源返回 `financial.dataset=\"nws_news\"`、`document_type=\"news\"`。它是新闻文档检索；不能据此宣称已接入完整公告、券商研报、财务指标、基金持仓或债券行情。客户端无需知道 RDS 表名，也不要自行添加 `filters.financial`、股票代码或指标等未在工具 schema 中定义的参数。\n- 当金融候选全部来自 NWS 新闻库时，服务在相关候选内按发布日期降序排序，同一时点再按检索相关性排序。靠前不代表更可信，也不保证是全库最新或完整的新闻集合。\n- 金融证据使用下文的字段与质量标记解释；需要验证关键数值或公司事件时，补查公告、交易所披露等原始材料，或从 `search_id` 启动研究。正文很长、有 URL 或检索评估为 `usable`，都不等于已读取并核验原文。\n\n### 解读搜索结果\n\n按以下顺序检查响应：\n\n1. 检查 `failures[]`，确认是否有来源或阶段失败。\n2. 检查 `result_set.counts_by_stage` 中 `web`、`academic`、`patent`、`legal`、`financial` 的 `recalled/ranked/assembled/selected` 计数，定位候选丢失阶段。\n3. 检查 `retrieval_assessment.status` 与 `retrieval_assessment.gaps[]`，判断证据是否足够。\n4. 检查 `query.filter_execution`，确认每项过滤器实际 `applied`、`unsupported` 或 `not_applicable` 的状态。\n5. 对法律证据读取 `type: \"legal\"`、`passage`、`legal.law_type`、`legal.status`、`legal.department`、`legal.directory` 与 `legal.item`。引用时保留法规标题和条号。\n6. 法律证据可能没有公开 URL，且 `access.is_open` 为 `false`。不要臆造官方链接、发布日期、法规效力或全文定位；原文被截断时遵守 `diagnostics` 中的限制。\n7. 金融证据读取 `type: \"financial\"`、`financial.dataset/document_type/vendor_record_id/tool`、`published_date`、`citation` 和 `passage.text`；检查 `quality.can_support_key_claim` 与 `diagnostics`。`PROVIDER_EXTRACT_NOT_ORIGINAL`、`NO_STABLE_LOCATOR` 或 `TRUNCATED_EVIDENCE` 表明摘录、定位或完整性限制，不能单独证明关键金融主张。缺少公开链接时保留标题、媒体、日期和记录 ID，不臆造 URL；有链接也不代表 `access.is_open=true`。\n8. 仅用 `evidence[].scores.relevance` 排序，不将其视为可信度或法律效力判断。\n\n`status` 只表示搜索执行是否完整，不代表证据充分性。`research_seed.search_id` 指向服务端保存的不可变 evidence 与检索边界快照；不要构造、修改或回传该快照。\n\n## `research`\n\n`research` 是持久化研究任务的统一生命周期工具，支持 `start`、`get`、`feedback` 和 `cancel`。\n\n### 启动任务\n\n从搜索响应取得 `research_seed.search_id` 后调用：\n\n```json\n{\n  \"operation\": \"start\",\n  \"search_id\": \"srch_...\",\n  \"idempotency_key\": \"agent-run-20260811-001\",\n  \"profile\": \"technology_validation\",\n  \"depth\": \"standard\",\n  \"objective\": {\n    \"question\": \"硫化物电解质的关键路线是什么，哪些已形成专利布局？\",\n    \"claims\": [\n      {\n        \"text\": \"硫化物电解质已形成较完整的专利布局\",\n        \"importance\": \"key\"\n      }\n    ],\n    \"required_features\": [\"离子电导率\", \"界面稳定性\", \"制备方法\"]\n  }\n}\n```\n\n- 必须提供 `search_id` 和全局 `idempotency_key`。为同一个逻辑请求的重试复用同一个 key；只有请求实质改变时才换 key。\n- 选择 `profile`：`literature_review`、`technology_validation`（默认）、`prior_art_landscape`、`technology_landscape` 或 `financial_research`；选择 `depth`：`quick`、`standard` 或 `deep`。\n- 把问题写入 `objective.question`，已知待核验陈述写入 `objective.claims`，必须覆盖的维度写入 `objective.required_features`。\n- 仅在要收紧预设上限时提供 `budget`；不要用它扩大预设预算。\n\n### 金融研究\n\n先检索目标公司取得真实 `search_id`，再启动金融研究，例如：\n\n```json\n{\n  \"operation\": \"start\",\n  \"search_id\": \"srch_...\",\n  \"idempotency_key\": \"financial-research-20260909-001\",\n  \"profile\": \"financial_research\",\n  \"depth\": \"standard\",\n  \"objective\": {\n    \"question\": \"查询贵州茅台（600519.SH）的最新成交价，并说明交易日期、交易时间和数据来源。\",\n    \"required_features\": [\"最新成交价\", \"交易日期与时间\", \"数据来源\"]\n  }\n}\n```\n\n- `financial_research` 是显式研究入口。服务端规划器可选择文档（`documents`）、股票行情（`price`）、K 线（`kline`）、基本面（`fundamentals`）、公司档案（`profile`）、公司事件（`events`）或 `unsupported`。这些是返回计划中的操作，不是客户端可直接调用的工具；股票代码、时间范围和所需指标写入 `objective.question`。\n- 结构化股票能力依赖服务端规划器和数据网关启用；不能把宁夏新闻搜索可用当作 Wind 股票数据已启用。当前结构化网关仅支持受限股票操作；行情与 K 线需明确 A 股代码（如 `600519.SH`），K 线还需日期区间和日/周/月周期。不要推断已支持基金、债券、宏观、全球股票或任意指标查询。\n- 读取 `resolved.financial_plan` 或最终 `dossier.financial_plan` 的 `intent_status`、`execution_status`、`operation`、`failure_code`。只有执行为 `ready` 且实际返回 `data` 时，才使用其结构化结果；`documents` 不执行股票数据查询。\n- 结构化结果位于 `financial_plan.data`，包括 `provider`、`tool`、`retrieved_at` 和 `tables[].columns/rows/units`；它们与 `evidence_index` 中的新闻文档分别引用。数值说明须保留证券主体、交易日或报告期、单位与来源；抓取时间不是行情时间，缺值不是零，不为表格臆造公开 URL。\n- 计划为 `disabled`、`failed` 或 `unsupported` 时，结合原因说明未取得哪些数据；可以使用仍可用的文档继续研究。`clarification_questions` 是待澄清信息，不能替用户补造证券代码、日期或口径。\n\n### 读取、补充与取消\n\n读取任务：\n\n```json\n{\n  \"operation\": \"get\",\n  \"research_id\": \"rsch_...\",\n  \"detail\": \"full\"\n}\n```\n\n- 当 `state` 为 `queued` 或 `running` 时，等待响应中的 `retry_after_ms` 再读取；不要高频轮询。\n- 仅在 `needs_input` 时提交 `feedback`。先读取最新任务，再以当前 `task_revision` 提交真实的 `answers` 或 `note`；不要替用户虚构答案。\n- 取消任务时使用最新的 `task_revision`。终态包括 `completed`、`partial`、`needs_input`、`failed` 与 `cancelled`。\n- `state=\"completed\"` 只表示流程正常停止。用 `dossier.assessment.overall` 判断结论充分性，并从 finding 的 evidence ID 解引用 `dossier.evidence_index`；同时检查 `dossier.coverage.gaps` 和顶层 `stop`。\n\n## 作答规则\n\n- 让每个关键结论对应具体 evidence，并保留 finding、evidence 和 locator 的关系。区分搜索片段、法规条文、PDF 原文和研究结论。\n- 法律问题先说明适用法域、检索时点、法条状态和证据缺口；将法条内容、事实适用和结论明确分开。对于个案、争议或跨法域问题，提示用户咨询合格的法律专业人士。\n- 学术结论优先使用可定位的原始论文；专利结论优先使用专利文献，并区分申请、公开、授权和未知状态。\n- 对时效性问题检查证据日期、`legal.status` 和 `filter_execution`。日期陈旧、状态待核实、过滤未执行或证据缺少公开链接时，明确披露限制。\n- 当 `retrieval_assessment.status` 为 `limited`，有实质 `failures[]`，或法律/研究证据为 `partial`、`insufficient`、`conflicted`、`needs_expert_review` 时，不要给出无保留的确定性结论。\n- 不要生成或声称接口提供单一 `trust_score`，不要臆造来源、法规元数据、引文、申请人、发明人、日期、许可证或 URL。\n\n## 错误处理\n\n- `401 invalid_token` 或未授权：重新执行 `openclaw mcp login chukonu-web-search`、`codex mcp login chukonu-web-search`，或在 Claude Code 中通过 `/mcp → chukonu-web-search → Authenticate` 重新完成 OAuth；同时检查并移除误配的静态 Authorization 头。\n- `search` 参数被拒绝：移除未知字段，确认 `limit`、`source_types`、日期和 `legal_status` 的格式。\n- 金融来源不可用或无结果：检查 `failures[]`、`query.filter_execution` 和日期范围；如改用网页补查，明确标明来源变化和金融数据缺口。`FINANCIAL_RESEARCH_PLANNER_DISABLED`、`WIND_STOCK_DATA_DISABLED` 等错误表示结构化能力未启用；不要把文档研究完成解释为行情查询成功，也不要要求用户配置上游凭据或绕过公开 MCP。\n- 法律 provider 不可用：阅读 `failures[]` 和 `retrieval_assessment.gaps[]`；只有在普通网页证据仍覆盖主张时才以明确限制继续作答。\n- `research start` 幂等冲突：若逻辑请求未变则恢复原请求并复用原 key；若请求变了才用新 key。\n- `research` 返回 `needs_input`：读取 `input_request`，向用户取得答案后用最新 revision 提交。返回 `failed`、`partial` 或 `cancelled`：读取 `failures[]`、`stop` 和 coverage gaps，并说明已完成部分与限制。\n\nFile v1.1.4:_meta.json\n\n{\n  \"ownerId\": \"kn7dq96kwex3nsqjg8bnbnz04h8a39jr\",\n  \"slug\": \"chukonu-web-search\",\n  \"version\": \"1.1.4\",\n  \"publishedAt\": 1788915746993\n}\n\nFile v1.1.4:skill-card.md\n\n## Description:\n\nUses the Chukonu remote MCP search and research tools to retrieve web, academic, patent, financial, and Chinese legal evidence for cited real-time research and verification tasks.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[hhhwor](https://clawhub.ai/user/hhhwor)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers and agents use this skill when they need externally sourced, current, citable evidence for search, fact checking, counter-evidence, Chinese legal retrieval, financial-news research, PDF review, and persistent multi-turn research workflows.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: Search queries and research objectives are sent to the Chukonu remote MCP service.\n\nMitigation: Avoid submitting sensitive information unless the user accepts that remote processing; use the normal OAuth flow and let the host manage tokens.\n\nRisk: Static Authorization headers, API tokens, or leaked OAuth material could create authentication and credential-handling risk.\n\nMitigation: Do not configure static Authorization headers or CHUKONU_SEARCH_API_TOKEN, and do not print or persist authorization codes, callback URLs, access tokens, or refresh tokens.\n\nRisk: Legal and financial results may have incomplete coverage, unsupported filters, missing public links, or source-specific limitations.\n\nMitigation: Check retrieval assessments, failures, filter execution, legal status, and financial data plans before making claims; present limitations and use primary or professional review for consequential decisions.\n\n## Reference(s):\n\n- [ClawHub skill page](https://clawhub.ai/hhhwor/skills/chukonu-web-search)\n- [Chukonu remote MCP endpoint](https://search.houdutech.cn/web/mcp/)\n\n## Skill Output:\n\n**Output Type(s):** [Text, Markdown, Shell commands, Configuration, Guidance]\n\n**Output Format:** [Markdown responses with citations, inline JSON examples, and shell command examples]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May include cited evidence, retrieval limitations, research status, and OAuth setup guidance.]\n\n## Skill Version(s):\n\n1.1.4 (source: server release evidence)\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\nArchive v1.1.3: 3 files, 5832 bytes\n\nFiles: skill-card.md (2405b), SKILL.md (8295b), _meta.json (137b)\n\nFile v1.1.3:SKILL.md\n\n---\nname: chukonu-web-search\ndescription: \"通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。\"\n---\n\n# Chukonu Search + Research\n\n通过 Chukonu remote MCP 调用公开的 `search` 和 `research` 工具。无需本地二进制、API key 或静态 token；不要调用内部工具、上游服务或内部端口。法律法规检索是服务内部接入的 FY MCP provider，客户端不要单独配置它、传入其凭据或尝试调用其工具。\n\n## 一次性接入\n\n使用 `https://search.houdutech.cn/web/mcp/` streamable-http 端点。该端点通过标准 OAuth 2.0 保护，采用动态客户端注册、授权码、PKCE(S256) 和 refresh token，scope 为 `search:read`。让 MCP host 自动发现并注册，不要手工填写 client ID 或 client secret。\n\nOpenClaw：\n\n```bash\nopenclaw mcp add chukonu-web-search \\\n  --url https://search.houdutech.cn/web/mcp/ \\\n  --transport streamable-http \\\n  --auth oauth\nopenclaw mcp login chukonu-web-search\n```\n\nClaude Code：\n\n```bash\nclaude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ \\\n  --scope user --transport http\n  \nCodex：\n\n```bash\ncodex mcp add chukonu-web-search --url https://search.houdutech.cn/web/mcp/\nodex mcp login chukonu-web-search\ncodex mcp list\n```\n\n# 然后在会话中：/mcp → chukonu-web-search → Authenticate\n```\n\n不要为该服务器配置静态 `Authorization` 头或 `CHUKONU_SEARCH_API_TOKEN`。静态 Authorization 头会使 host 禁用 OAuth 回退，并可能导致 `401 invalid_token`。把 OAuth token 完全交给 host 保管，不要在日志、回答、示例或错误信息中泄露。\n\n## 选择工具与来源\n\n- 用 `search` 处理秒级发现、法规定位和获取后续研究所需的 `research_seed.search_id`。\n- 用 `research` 处理关键主张核验、反证、跨来源冲突、全文定位或证据覆盖评估。必须先 `search`，再用返回的 `search_id` 启动任务。\n- `source_types` 只接受 `web`、`academic`、`patent` 和 `legal`。省略时服务自动路由。、\n\n## `search`\n\n### 请求\n\n最小调用：\n\n```json\n{\n  \"query\": \"固态电池硫化物电解质近五年的关键路线\"\n}\n```\n\n需要约束来源或过滤范围时：\n\n```json\n{\n  \"query\": \"固态电池硫化物电解质近五年的关键路线\",\n  \"limit\": 10,\n  \"source_types\": [\"web\", \"academic\", \"patent\"],\n  \"filters\": {\n    \"published_from\": \"2021-01-01\",\n    \"published_to\": \"2026-07-17\",\n    \"languages\": [\"zh\", \"en\"],\n    \"jurisdictions\": [\"CN\", \"US\", \"EP\", \"WO\"]\n  }\n}\n```\n\n遵守以下约束：\n\n- `query` 必填；保持查询简洁、具体。\n- `limit` 为最终全局返回数，范围是 1–20。\n- `source_types` 仅接受 `web`、`academic`、`patent`。省略时让服务自动路由；传入后只检索指定类型。\n- `filters` 支持 `published_from`、`published_to`、`languages`、`jurisdictions`。\n- 不要发送请求级模型、重排、PDF 或 trust 开关。未知字段会被拒绝。\n\n### 解读搜索结果\n\n按以下顺序检查响应：\n\n1. 检查 `failures[]`，确认是否有来源或阶段失败。\n2. 检查 `result_set.counts_by_stage` 中 `web`、`academic`、`patent`、`legal` 的 `recalled/ranked/assembled/selected` 计数，定位候选丢失阶段。\n3. 检查 `retrieval_assessment.status` 与 `retrieval_assessment.gaps[]`，判断证据是否足够。\n4. 检查 `query.filter_execution`，确认每项过滤器实际 `applied`、`unsupported` 或 `not_applicable` 的状态。\n5. 对法律证据读取 `type: \"legal\"`、`passage`、`legal.law_type`、`legal.status`、`legal.department`、`legal.directory` 与 `legal.item`。引用时保留法规标题和条号。\n6. 法律证据可能没有公开 URL，且 `access.is_open` 为 `false`。不要臆造官方链接、发布日期、法规效力或全文定位；原文被截断时遵守 `diagnostics` 中的限制。\n7. 仅用 `evidence[].scores.relevance` 排序，不将其视为可信度或法律效力判断。\n\n`status` 只表示搜索执行是否完整，不代表证据充分性。`research_seed.search_id` 指向服务端保存的不可变 evidence 与检索边界快照；不要构造、修改或回传该快照。\n\n## `research`\n\n`research` 是持久化研究任务的统一生命周期工具，支持 `start`、`get`、`feedback` 和 `cancel`。\n\n### 启动任务\n\n从搜索响应取得 `research_seed.search_id` 后调用：\n\n```json\n{\n  \"operation\": \"start\",\n  \"search_id\": \"srch_...\",\n  \"idempotency_key\": \"agent-run-20260811-001\",\n  \"profile\": \"technology_validation\",\n  \"depth\": \"standard\",\n  \"objective\": {\n    \"question\": \"硫化物电解质的关键路线是什么，哪些已形成专利布局？\",\n    \"claims\": [\n      {\n        \"text\": \"硫化物电解质已形成较完整的专利布局\",\n        \"importance\": \"key\"\n      }\n    ],\n    \"required_features\": [\"离子电导率\", \"界面稳定性\", \"制备方法\"]\n  }\n}\n```\n\n- 必须提供 `search_id` 和全局 `idempotency_key`。为同一个逻辑请求的重试复用同一个 key；只有请求实质改变时才换 key。\n- 选择 `profile`：`literature_review`、`technology_validation`（默认）、`prior_art_landscape` 或 `technology_landscape`；选择 `depth`：`quick`、`standard` 或 `deep`。\n- 把问题写入 `objective.question`，已知待核验陈述写入 `objective.claims`，必须覆盖的维度写入 `objective.required_features`。\n- 仅在要收紧预设上限时提供 `budget`；不要用它扩大预设预算。\n\n### 读取、补充与取消\n\n读取任务：\n\n```json\n{\n  \"operation\": \"get\",\n  \"research_id\": \"rsch_...\",\n  \"detail\": \"full\"\n}\n```\n\n- 当 `state` 为 `queued` 或 `running` 时，等待响应中的 `retry_after_ms` 再读取；不要高频轮询。\n- 仅在 `needs_input` 时提交 `feedback`。先读取最新任务，再以当前 `task_revision` 提交真实的 `answers` 或 `note`；不要替用户虚构答案。\n- 取消任务时使用最新的 `task_revision`。终态包括 `completed`、`partial`、`needs_input`、`failed` 与 `cancelled`。\n- `state=\"completed\"` 只表示流程正常停止。用 `dossier.assessment.overall` 判断结论充分性，并从 finding 的 evidence ID 解引用 `dossier.evidence_index`；同时检查 `dossier.coverage.gaps` 和顶层 `stop`。\n\n## 作答规则\n\n- 让每个关键结论对应具体 evidence，并保留 finding、evidence 和 locator 的关系。区分搜索片段、法规条文、PDF 原文和研究结论。\n- 法律问题先说明适用法域、检索时点、法条状态和证据缺口；将法条内容、事实适用和结论明确分开。对于个案、争议或跨法域问题，提示用户咨询合格的法律专业人士。\n- 学术结论优先使用可定位的原始论文；专利结论优先使用专利文献，并区分申请、公开、授权和未知状态。\n- 对时效性问题检查证据日期、`legal.status` 和 `filter_execution`。日期陈旧、状态待核实、过滤未执行或证据缺少公开链接时，明确披露限制。\n- 当 `retrieval_assessment.status` 为 `limited`，有实质 `failures[]`，或法律/研究证据为 `partial`、`insufficient`、`conflicted`、`needs_expert_review` 时，不要给出无保留的确定性结论。\n- 不要生成或声称接口提供单一 `trust_score`，不要臆造来源、法规元数据、引文、申请人、发明人、日期、许可证或 URL。\n\n## 错误处理\n\n- `401 invalid_token` 或未授权：重新执行 `openclaw mcp login chukonu-web-search`，或在 Claude Code 中通过 `/mcp → chukonu-web-search → Authenticate` 重新完成 OAuth；同时检查并移除误配的静态 Authorization 头。\n- `search` 参数被拒绝：移除未知字段，确认 `limit`、`source_types`、日期和 `legal_status` 的格式。\n- `research` 返回 `needs_input`：读取 `input_request`，向用户取得答案后用最新 revision 提交。返回 `failed`、`partial` 或 `cancelled`：读取 `failures[]`、`stop` 和 coverage gaps，并说明已完成部分与限制。\n\nFile v1.1.3:_meta.json\n\n{\n  \"ownerId\": \"kn7dq96kwex3nsqjg8bnbnz04h8a39jr\",\n  \"slug\": \"chukonu-web-search\",\n  \"version\": \"1.1.3\",\n  \"publishedAt\": 1786412847497\n}\n\nFile v1.1.3:skill-card.md\n\n## Description:\n\nGuides agents in using the Chukonu remote MCP search and research tools to retrieve web, academic, patent, and Chinese legal or regulatory evidence for cited answers, fact checks, prior-art review, PDF-focused research, and multi-turn research tasks.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[hhhwor](https://clawhub.ai/user/hhhwor)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers and agent users use this skill to configure OAuth access to the Chukonu search MCP and to guide search, research, citation, legal evidence handling, and result-limitation disclosure workflows.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: Search queries and research objectives are sent to a third-party remote MCP service.\n\nMitigation: Confirm trust in the Chukonu service before installation and avoid submitting sensitive information unless the deployment policy permits it.\n\nRisk: Misconfigured static authorization headers or exposed tokens can cause authentication failures or credential leakage.\n\nMitigation: Use the documented OAuth flow with host-managed tokens, and do not place OAuth tokens in logs, examples, or manual Authorization headers.\n\nRisk: Retrieved evidence, especially legal or time-sensitive material, may be partial, unavailable by public URL, stale, or insufficient for a definitive conclusion.\n\nMitigation: Check retrieval assessments, failures, legal status fields, dates, and coverage gaps, and disclose limitations before presenting conclusions.\n\n## Reference(s):\n\n- [ClawHub skill page](https://clawhub.ai/hhhwor/skills/chukonu-web-search)\n- [Chukonu remote MCP endpoint](https://search.houdutech.cn/web/mcp/)\n\n## Skill Output:\n\n**Output Type(s):** [text, markdown, shell commands, configuration, guidance]\n\n**Output Format:** [Markdown with inline JSON and shell command examples]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [The skill guides remote MCP calls and evidence-grounded answers; it does not produce local executable payloads.]\n\n## Skill Version(s):\n\n1.1.3 (source: server release evidence)\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\nArchive v1.1.1: 3 files, 6257 bytes\n\nFiles: skill-card.md (2600b), SKILL.md (9566b), _meta.json (137b)\n\nFile v1.1.1:SKILL.md\n\n---\nname: chukonu-web-search\ndescription: \"通过 Chukonu remote MCP 的 search 与 research 获取网页、学术和专利证据。用于需要外部、实时、可引用信息的搜索任务，以及需要事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究的任务；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。\"\nversion: 0.3.1\nmetadata:\n  version: \"0.3.1\"\n  tags: [web, search, research, academic, patent, mcp, evidence, oauth]\n  transport: streamable-http\n  requires:\n    mcp:\n      - name: chukonu-web-search\n        url: \"https://search.houdutech.cn/web/mcp/\"\n        transport: streamable-http\n        auth: oauth\n  setup:\n    - \"openclaw mcp add chukonu-web-search --url https://search.houdutech.cn/web/mcp/ --transport streamable-http --auth oauth\"\n    - \"openclaw mcp login chukonu-web-search\"\n    - \"claude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ --scope user --transport http\"\n    - \"在 Claude Code 中：/mcp → 选择 chukonu-web-search → Authenticate 完成浏览器 OAuth\"\n---\n\n# Chukonu Search + Research\n\n通过 Chukonu remote MCP 获取网页、学术和专利证据。本 skill 不调用本地二进制，也不需要 API key 或静态 token。只使用公开的 `search` 与 `research` 两个业务工具；陈述校验和 PDF 全文读取是 `research` 的内部阶段，不要调用 `verify_claims`、`get_pdf_text` 或内部上游服务，也不要直连 9001、9101 等内部端口。\n\n## 一次性接入\n\n使用 `https://search.houdutech.cn/web/mcp/` streamable-http 端点。该端点通过标准 OAuth 2.0 保护，采用动态客户端注册、授权码、PKCE(S256) 和 refresh token，scope 为 `search:read`。让 MCP host 自动发现并注册，不要手工填写 client ID 或 client secret。\n\nOpenClaw：\n\n```bash\nopenclaw mcp add chukonu-web-search \\\n  --url https://search.houdutech.cn/web/mcp/ \\\n  --transport streamable-http \\\n  --auth oauth\nopenclaw mcp login chukonu-web-search\n```\n\nClaude Code：\n\n```bash\nclaude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ \\\n  --scope user --transport http\n# 然后在会话中：/mcp → chukonu-web-search → Authenticate\n```\n\n不要为该服务器配置静态 `Authorization` 头或 `CHUKONU_SEARCH_API_TOKEN`。静态 Authorization 头会使 host 禁用 OAuth 回退，并可能导致 `401 invalid_token`。把 OAuth token 完全交给 host 保管，不要在日志、回答、示例或错误信息中泄露。\n\n## 选择工具\n\n- 用 `search` 完成秒级、单轮、轻量发现，以及获取后续研究所需的 `search_id`。\n- 当任务需要事实核验、反证、PDF 全文、证据覆盖评估或可信的多轮研究时，先调用 `search`，再用其 `research_seed.search_id` 启动 `research`。\n- 不要把搜索结果的相关性排序当作事实置信度。高风险结论或证据存在明显缺口时，优先转入 `research`。\n\n## `search`\n\n### 请求\n\n最小调用：\n\n```json\n{\n  \"query\": \"固态电池硫化物电解质近五年的关键路线\"\n}\n```\n\n需要约束来源或过滤范围时：\n\n```json\n{\n  \"query\": \"固态电池硫化物电解质近五年的关键路线\",\n  \"limit\": 10,\n  \"source_types\": [\"web\", \"academic\", \"patent\"],\n  \"filters\": {\n    \"published_from\": \"2021-01-01\",\n    \"published_to\": \"2026-07-17\",\n    \"languages\": [\"zh\", \"en\"],\n    \"jurisdictions\": [\"CN\", \"US\", \"EP\", \"WO\"]\n  }\n}\n```\n\n遵守以下约束：\n\n- `query` 必填；保持查询简洁、具体。\n- `limit` 为最终全局返回数，范围是 1–20。\n- `source_types` 仅接受 `web`、`academic`、`patent`。省略时让服务自动路由；传入后只检索指定类型。\n- `filters` 支持 `published_from`、`published_to`、`languages`、`jurisdictions`。\n- 不要发送请求级模型、重排、PDF 或 trust 开关。未知字段会被拒绝。\n\n### 检查结果\n\n按以下顺序检查搜索响应：\n\n1. 读取 `failures[]`，确认是否有来源或阶段失败。\n2. 读取 `result_set.counts_by_stage`，比较 `recalled/ranked/assembled/selected` 的分来源计数，定位候选在哪个阶段损失。\n3. 读取 `retrieval_assessment.status` 和 `retrieval_assessment.gaps[]`，判断证据是否可用及缺口。\n4. 读取 `query.filter_execution`，确认所需过滤器是否真正应用；不要仅根据请求参数假设过滤成功。\n5. 检查每条 `evidence[].quality`，优先使用质量更高、可定位、可引用的证据。\n6. 仅把 `evidence[].scores.relevance` 用于排序，不要将其解释为事实置信度。\n\n`status` 只表示搜索执行是否完整，不能代替证据充分性判断。`research_seed.search_id` 指向服务端保存的不可变 evidence 与检索边界快照；不要自行构造、修改或让客户端回传该快照。\n\n## `research`\n\n`research` 是持久化研究任务的统一生命周期工具，支持 `start`、`get`、`feedback` 和 `cancel`。\n\n### 启动任务\n\n从搜索响应取得 `research_seed.search_id` 后调用：\n\n```json\n{\n  \"operation\": \"start\",\n  \"search_id\": \"srch_...\",\n  \"idempotency_key\": \"agent-run-20260717-001\",\n  \"profile\": \"technology_validation\",\n  \"depth\": \"standard\",\n  \"objective\": {\n    \"question\": \"硫化物电解质的关键路线是什么，哪些已形成专利布局？\",\n    \"claims\": [\n      {\n        \"text\": \"硫化物电解质已形成较完整的专利布局\",\n        \"importance\": \"key\"\n      }\n    ],\n    \"required_features\": [\"离子电导率\", \"界面稳定性\", \"制备方法\"]\n  }\n}\n```\n\n启动时：\n\n- 必须提供 `search_id` 和全局 `idempotency_key`。\n- 为同一个逻辑请求的重试复用同一个幂等键。相同 key 与等价请求返回同一任务；相同 key 用于不同请求会失败。\n- 根据目标选择 `profile`：`literature_review`、`technology_validation`（默认）、`prior_art_landscape` 或 `technology_landscape`。\n- 根据研究强度选择 `depth`：`quick`、`standard` 或 `deep`。\n- 仅在需要收紧预设上限时显式传入 `budget`；不要用它扩大预设预算。\n- 用 `objective.question` 表达问题；已知待核验陈述放入 `objective.claims`；必须覆盖的维度放入 `objective.required_features`。\n\n### 读取与轮询\n\n```json\n{\n  \"operation\": \"get\",\n  \"research_id\": \"rsch_...\",\n  \"detail\": \"full\"\n}\n```\n\n当 `state` 为 `queued` 或 `running` 时，等待响应给出的 `retry_after_ms` 后再次读取，不要高频轮询。终态包括 `completed`、`partial`、`needs_input`、`failed` 和 `cancelled`。\n\n### 补充输入\n\n仅当任务处于 `needs_input` 时提交 feedback。先读取最新任务，使用其当前 `task_revision`：\n\n```json\n{\n  \"operation\": \"feedback\",\n  \"research_id\": \"rsch_...\",\n  \"task_revision\": 3,\n  \"answers\": {\n    \"target\": \"只关注量产路线\"\n  }\n}\n```\n\n需要用户选择时，先把 `input_request` 中的问题转述给用户；不要替用户虚构答案。`task_revision` 用于并发保护，发生版本冲突时重新读取任务后再决定是否提交。\n\n### 取消任务\n\n```json\n{\n  \"operation\": \"cancel\",\n  \"research_id\": \"rsch_...\",\n  \"task_revision\": 3\n}\n```\n\n取消前使用最新的 `task_revision`，避免基于过期状态操作任务。\n\n## 解读研究结果\n\n- `state=\"completed\"` 只表示研究流程正常停止，不表示结论已被证实。\n- 用 `dossier.assessment.overall` 判断结论充分性：`sufficient`、`sufficient_with_limitations`、`insufficient`、`conflicted` 或 `needs_expert_review`。\n- 从每个 finding 的 evidence 引用 ID 出发，在 `dossier.evidence_index` 中解引用原始证据。\n- 从 `dossier.coverage.gaps` 读取尚未覆盖的问题，从顶层 `stop` 读取停止原因。\n- 不要生成或声称接口返回了单一 `trust_score`。\n\n## 作答规则\n\n- 让关键论断落到具体 evidence；不要臆造来源、元数据、引文、申请人、发明人、日期、许可或专利状态。\n- 学术结论优先使用可定位的原始论文证据；专利结论优先使用专利文献，并区分申请、公开、授权及未知状态。\n- 对时效性问题检查证据日期；日期陈旧、缺失或过滤未落实时明确说明限制。\n- 搜索为 `limited` 或存在实质性 `failures[]` 时，可以使用仍有效的证据，但必须披露会影响结论的缺口；需要核验时转入 `research`。\n- 研究为 `partial`、`insufficient`、`conflicted` 或 `needs_expert_review` 时，不要输出无保留的确定性结论。\n- 引用研究结果时同时保留 finding、evidence 和 locator 之间的对应关系，不要把摘要、搜索片段和 PDF 原文混为一谈。\n\n## 错误与边界处理\n\n- `401 invalid_token` 或未授权：重新执行 `openclaw mcp login chukonu-web-search`，或在 Claude Code 中通过 `/mcp → chukonu-web-search → Authenticate` 重新完成 OAuth；同时检查并移除误配的静态 Authorization 头。\n- `search` 参数被拒绝：移除未知字段，并检查 `limit`、来源类型、日期和过滤器格式。\n- `research start` 幂等冲突：如果逻辑请求未变，恢复原请求并复用原 key；只有请求确实改变时才使用新 key。\n- `research` 返回 `needs_input`：读取 `input_request`，获得真实答案后用最新 revision 提交 feedback。\n- `research` 返回 `failed`、`partial` 或 `cancelled`：读取 `failures[]`、`stop` 和 coverage gaps，向用户说明已完成部分及限制。\n\n远程 MCP 的鉴权由 host 按 OAuth 流程完成。服务不建立租户身份模型；始终把 OAuth 凭据视为秘密。\n\nFile v1.1.1:_meta.json\n\n{\n  \"ownerId\": \"kn7dq96kwex3nsqjg8bnbnz04h8a39jr\",\n  \"slug\": \"chukonu-web-search\",\n  \"version\": \"1.1.1\",\n  \"publishedAt\": 1785804417935\n}\n\nFile v1.1.1:skill-card.md\n\n## Description: <br>\nChukonu Web Search helps agents use the Chukonu remote MCP service to retrieve web, academic, and patent evidence and run deeper research for claim checking, PDF review, coverage assessment, and multi-turn investigations. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[hhhwor](https://clawhub.ai/user/hhhwor) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and external users use this skill when an agent needs current, citable search evidence, source-quality checks, patent or academic discovery, or a persisted research dossier before answering. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: Search queries, claims, and research objectives are sent to the remote Chukonu MCP service after OAuth authentication. <br>\nMitigation: Use the skill only where that remote service is approved for the data involved, and avoid putting secrets or private data into research prompts. <br>\nRisk: Current search and research results can be incomplete, stale, or affected by retrieval gaps. <br>\nMitigation: Review failures, retrieval assessments, coverage gaps, evidence quality, and citation locators before relying on conclusions. <br>\nRisk: OAuth tokens are sensitive credentials handled by the MCP host. <br>\nMitigation: Do not configure static authorization headers or expose OAuth tokens in logs, answers, examples, or error messages. <br>\n\n\n## Reference(s): <br>\n- [ClawHub skill page](https://clawhub.ai/hhhwor/skills/chukonu-web-search) <br>\n- [Publisher profile](https://clawhub.ai/user/hhhwor) <br>\n- [Chukonu MCP endpoint](https://search.houdutech.cn/web/mcp/) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Guidance, Shell commands, Configuration, Text, Markdown] <br>\n**Output Format:** [Markdown guidance with inline shell commands, JSON request examples, and evidence-backed response instructions] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [Requires OAuth authentication to the remote Chukonu MCP service; search and research outputs should preserve evidence, retrieval assessment, and citation relationships.] <br>\n\n## Skill Version(s): <br>\n1.1.1 (source: server release metadata; artifact frontmatter metadata version 0.3.1) <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 v1.1.0: 3 files, 6155 bytes\n\nFiles: skill-card.md (2370b), SKILL.md (9526b), _meta.json (137b)\n\nFile v1.1.0:SKILL.md\n\n---\nname: chukonu-web-search\ndescription: \"通过 Chukonu remote MCP 的 search 与 research 获取网页、学术和专利证据。用于需要外部、实时、可引用信息的搜索任务，以及需要事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究的任务；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。\"\nmetadata:\n  version: \"0.3.0\"\n  tags: [web, search, research, academic, patent, mcp, evidence, oauth]\n  transport: streamable-http\n  requires:\n    mcp:\n      - name: chukonu-web-search\n        url: \"https://search.houdutech.cn/web/mcp/\"\n        transport: streamable-http\n        auth: oauth\n  setup:\n    - \"openclaw mcp add chukonu-web-search --url https://search.houdutech.cn/web/mcp/ --transport streamable-http --auth oauth\"\n    - \"openclaw mcp login chukonu-web-search\"\n    - \"claude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ --scope user --transport http\"\n    - \"在 Claude Code 中：/mcp → 选择 chukonu-web-search → Authenticate 完成浏览器 OAuth\"\n---\n\n# Chukonu Search + Research\n\n通过 Chukonu remote MCP 获取网页、学术和专利证据。本 skill 不调用本地二进制，也不需要 API key 或静态 token。只使用公开的 `search` 与 `research` 两个业务工具；陈述校验和 PDF 全文读取是 `research` 的内部阶段，不要调用 `verify_claims`、`get_pdf_text` 或内部上游服务。\n\n## 一次性接入\n\n使用 `https://search.houdutech.cn/web/mcp/` streamable-http 端点。该端点通过标准 OAuth 2.0 保护，采用动态客户端注册、授权码、PKCE(S256) 和 refresh token，scope 为 `search:read`。让 MCP host 自动发现并注册，不要手工填写 client ID 或 client secret。\n\nOpenClaw：\n\n```bash\nopenclaw mcp add chukonu-web-search \\\n  --url https://search.houdutech.cn/web/mcp/ \\\n  --transport streamable-http \\\n  --auth oauth\nopenclaw mcp login chukonu-web-search\n```\n\nClaude Code：\n\n```bash\nclaude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ \\\n  --scope user --transport http\n# 然后在会话中：/mcp → chukonu-web-search → Authenticate\n```\n\n不要为该服务器配置静态 `Authorization` 头或 `CHUKONU_SEARCH_API_TOKEN`。静态 Authorization 头会使 host 禁用 OAuth 回退，并可能导致 `401 invalid_token`。把 OAuth token 完全交给 host 保管，不要在日志、回答、示例或错误信息中泄露。\n\n## 选择工具\n\n- 用 `search` 完成秒级、单轮、轻量发现，以及获取后续研究所需的 `search_id`。\n- 当任务需要事实核验、反证、PDF 全文、证据覆盖评估或可信的多轮研究时，先调用 `search`，再用其 `research_seed.search_id` 启动 `research`。\n- 不要把搜索结果的相关性排序当作事实置信度。高风险结论或证据存在明显缺口时，优先转入 `research`。\n\n## `search`\n\n### 请求\n\n最小调用：\n\n```json\n{\n  \"query\": \"固态电池硫化物电解质近五年的关键路线\"\n}\n```\n\n需要约束来源或过滤范围时：\n\n```json\n{\n  \"query\": \"固态电池硫化物电解质近五年的关键路线\",\n  \"limit\": 10,\n  \"source_types\": [\"web\", \"academic\", \"patent\"],\n  \"filters\": {\n    \"published_from\": \"2021-01-01\",\n    \"published_to\": \"2026-07-17\",\n    \"languages\": [\"zh\", \"en\"],\n    \"jurisdictions\": [\"CN\", \"US\", \"EP\", \"WO\"]\n  }\n}\n```\n\n遵守以下约束：\n\n- `query` 必填；保持查询简洁、具体。\n- `limit` 为最终全局返回数，范围是 1–20。\n- `source_types` 仅接受 `web`、`academic`、`patent`。省略时让服务自动路由；传入后只检索指定类型。\n- `filters` 支持 `published_from`、`published_to`、`languages`、`jurisdictions`。\n- 不要发送请求级模型、重排、PDF 或 trust 开关。未知字段会被拒绝。\n\n### 检查结果\n\n按以下顺序检查搜索响应：\n\n1. 读取 `failures[]`，确认是否有来源或阶段失败。\n2. 读取 `retrieval_assessment.status` 和 `retrieval_assessment.gaps[]`，判断证据是否可用及缺口。\n3. 读取 `query.filter_execution`，确认所需过滤器是否真正应用；不要仅根据请求参数假设过滤成功。\n4. 检查每条 `evidence[].quality`，优先使用质量更高、可定位、可引用的证据。\n5. 仅把 `evidence[].scores.relevance` 用于排序，不要将其解释为事实置信度。\n\n`status` 只表示搜索执行是否完整，不能代替证据充分性判断。`research_seed.search_id` 指向服务端保存的不可变 evidence 与检索边界快照；不要自行构造、修改或让客户端回传该快照。\n\n## `research`\n\n`research` 是持久化研究任务的统一生命周期工具，支持 `start`、`get`、`feedback` 和 `cancel`。\n\n### 启动任务\n\n从搜索响应取得 `research_seed.search_id` 后调用：\n\n```json\n{\n  \"operation\": \"start\",\n  \"search_id\": \"srch_...\",\n  \"idempotency_key\": \"agent-run-20260717-001\",\n  \"profile\": \"technology_validation\",\n  \"depth\": \"standard\",\n  \"objective\": {\n    \"question\": \"硫化物电解质的关键路线是什么，哪些已形成专利布局？\",\n    \"claims\": [\n      {\n        \"text\": \"硫化物电解质已形成较完整的专利布局\",\n        \"importance\": \"key\"\n      }\n    ],\n    \"required_features\": [\"离子电导率\", \"界面稳定性\", \"制备方法\"]\n  }\n}\n```\n\n启动时：\n\n- 必须提供 `search_id` 和全局 `idempotency_key`。\n- 为同一个逻辑请求的重试复用同一个幂等键。相同 key 与等价请求返回同一任务；相同 key 用于不同请求会失败。\n- 根据目标选择 `profile`：`literature_review`、`technology_validation`（默认）、`prior_art_landscape` 或 `technology_landscape`。\n- 根据研究强度选择 `depth`：`quick`、`standard` 或 `deep`。\n- 仅在需要收紧预设上限时显式传入 `budget`；不要用它扩大预设预算。\n- 用 `objective.question` 表达问题；已知待核验陈述放入 `objective.claims`；必须覆盖的维度放入 `objective.required_features`。\n\n工具另外暴露若干可选高级参数：`note`、`policy`、`privacy`、`scope`。默认省略；仅在明确了解其语义、且确有需要时才传入。\n\n### 读取与轮询\n\n```json\n{\n  \"operation\": \"get\",\n  \"research_id\": \"rsch_...\",\n  \"detail\": \"full\"\n}\n```\n\n当 `state` 为 `queued` 或 `running` 时，等待响应给出的 `retry_after_ms` 后再次读取，不要高频轮询。终态包括 `completed`、`partial`、`needs_input`、`failed` 和 `cancelled`。\n\n### 补充输入\n\n仅当任务处于 `needs_input` 时提交 feedback。先读取最新任务，使用其当前 `task_revision`：\n\n```json\n{\n  \"operation\": \"feedback\",\n  \"research_id\": \"rsch_...\",\n  \"task_revision\": 3,\n  \"answers\": {\n    \"target\": \"只关注量产路线\"\n  }\n}\n```\n\n需要用户选择时，先把 `input_request` 中的问题转述给用户；不要替用户虚构答案。`task_revision` 用于并发保护，发生版本冲突时重新读取任务后再决定是否提交。\n\n### 取消任务\n\n```json\n{\n  \"operation\": \"cancel\",\n  \"research_id\": \"rsch_...\",\n  \"task_revision\": 3\n}\n```\n\n取消前使用最新的 `task_revision`，避免基于过期状态操作任务。\n\n## 解读研究结果\n\n- `state=\"completed\"` 只表示研究流程正常停止，不表示结论已被证实。\n- 用 `dossier.assessment.overall` 判断结论充分性：`sufficient`、`sufficient_with_limitations`、`insufficient`、`conflicted` 或 `needs_expert_review`。\n- 从每个 finding 的 evidence 引用 ID 出发，在 `dossier.evidence_index` 中解引用原始证据。\n- 从 `dossier.coverage.gaps` 读取尚未覆盖的问题，从顶层 `stop` 读取停止原因。\n- 不要生成或声称接口返回了单一 `trust_score`。\n\n## 作答规则\n\n- 让关键论断落到具体 evidence；不要臆造来源、元数据、引文、申请人、发明人、日期、许可或专利状态。\n- 学术结论优先使用可定位的原始论文证据；专利结论优先使用专利文献，并区分申请、公开、授权及未知状态。\n- 对时效性问题检查证据日期；日期陈旧、缺失或过滤未落实时明确说明限制。\n- 搜索为 `limited` 或存在实质性 `failures[]` 时，可以使用仍有效的证据，但必须披露会影响结论的缺口；需要核验时转入 `research`。\n- 研究为 `partial`、`insufficient`、`conflicted` 或 `needs_expert_review` 时，不要输出无保留的确定性结论。\n- 引用研究结果时同时保留 finding、evidence 和 locator 之间的对应关系，不要把摘要、搜索片段和 PDF 原文混为一谈。\n\n## 错误与边界处理\n\n- `401 invalid_token` 或未授权：重新执行 `openclaw mcp login chukonu-web-search`，或在 Claude Code 中通过 `/mcp → chukonu-web-search → Authenticate` 重新完成 OAuth；同时检查并移除误配的静态 Authorization 头。\n- `search` 参数被拒绝：移除未知字段，并检查 `limit`、来源类型、日期和过滤器格式。\n- `research start` 幂等冲突：如果逻辑请求未变，恢复原请求并复用原 key；只有请求确实改变时才使用新 key。\n- `research` 返回 `needs_input`：读取 `input_request`，获得真实答案后用最新 revision 提交 feedback。\n- `research` 返回 `failed`、`partial` 或 `cancelled`：读取 `failures[]`、`stop` 和 coverage gaps，向用户说明已完成部分及限制。\n\n远程 MCP 的鉴权由 host 按 OAuth 流程完成。服务不建立租户身份模型；始终把 OAuth 凭据视为秘密。\n\nFile v1.1.0:_meta.json\n\n{\n  \"ownerId\": \"kn7dq96kwex3nsqjg8bnbnz04h8a39jr\",\n  \"slug\": \"chukonu-web-search\",\n  \"version\": \"1.1.0\",\n  \"publishedAt\": 1784601484594\n}\n\nFile v1.1.0:skill-card.md\n\n## Description: <br>\nUses the Chukonu remote MCP service to search web, academic, and patent sources, run deeper research tasks, and answer with structured evidence, retrieval assessments, and research dossiers. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[hhhwor](https://clawhub.ai/user/hhhwor) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and research-oriented agents use this skill to run current web, academic, and patent searches through a remote MCP service, then escalate to persistent research when the task needs fact checking, counterevidence, PDF review, or coverage assessment. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: Search queries, research objectives, and related context are sent to the Chukonu MCP service. <br>\nMitigation: Avoid using the skill for confidential research unless the provider's privacy terms meet the user's needs. <br>\nRisk: OAuth credentials are stored and managed by the MCP host. <br>\nMitigation: Use the host's OAuth flow, do not configure static Authorization headers, and avoid exposing tokens in logs, examples, or answers. <br>\nRisk: Search rankings and completed research runs can be mistaken for factual certainty. <br>\nMitigation: Base conclusions on cited evidence, retrieval assessments, coverage gaps, and dossier assessments rather than relevance scores or task completion state alone. <br>\n\n\n## Reference(s): <br>\n- [ClawHub Skill Page](https://clawhub.ai/hhhwor/skills/chukonu-web-search) <br>\n- [Chukonu Web Search MCP Endpoint](https://search.houdutech.cn/web/mcp/) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [text, markdown, shell commands, configuration, guidance] <br>\n**Output Format:** [Markdown with inline JSON and bash examples] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [Evidence-grounded responses should preserve links between findings, evidence IDs, and source locators.] <br>\n\n## Skill Version(s): <br>\n1.1.0 (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 v1.0.2: 3 files, 5819 bytes\n\nFiles: skill-card.md (2182b), SKILL.md (9461b), _meta.json (137b)\n\nFile v1.0.2:SKILL.md\n\n---\nname: chukonu-web-search\nversion: 0.2.0\ndescription: \"网络搜索：通过 Chukonu remote MCP server 获取面向 Agent 的网页、学术与专利检索证据。自然语言查询，返回证据优先的结构化结果（evidence[]、answerability 可答性、partial_failure 部分失败诊断），可选学术/专利意图开关与重排。只读、OAuth 登录。\"\ntags: [web, search, research, academic, patent, mcp, evidence, oauth]\nmetadata:\n  transport: streamable-http\n  requires:\n    mcp:\n      - name: chukonu-web-search\n        url: \"https://search.houdutech.cn/web/mcp/\"\n        transport: streamable-http\n        auth: oauth\n  setup:\n    # OpenClaw\n    - \"openclaw mcp add chukonu-web-search --url https://search.houdutech.cn/web/mcp/ --transport streamable-http --auth oauth\"\n    - \"openclaw mcp login chukonu-web-search\"\n    # Claude Code（本次实践；切勿配置静态 Authorization 头，否则会禁用 OAuth）\n    - \"claude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ --scope user --transport http\"\n    - \"在 Claude Code 中：/mcp → 选择 chukonu-web-search → Authenticate 完成浏览器 OAuth\"\n---\n\n# chukonu-web-search\n\n通过 Chukonu 的 **remote MCP server** 检索网页、学术与专利证据。本 skill 不调用本地二进制，也**不需要 API key / 静态 token** —— 它把 host（OpenClaw 或 Claude Code）接到托管的 `chukonu-web-search` MCP server，**OAuth 登录、只读检索**。\n\n## 一次性接入（首次使用前）\n\n端点：`https://search.houdutech.cn/web/mcp/`（streamable-http）。该端点通过标准 OAuth 2.0 保护：动态客户端注册 + 授权码 + PKCE(S256) + refresh_token，scope `search:read`。host 会自动发现并注册，无需手工填 client_id/secret。\n\nOpenClaw：\n\n```bash\nopenclaw mcp add chukonu-web-search \\\n  --url https://search.houdutech.cn/web/mcp/ \\\n  --transport streamable-http \\\n  --auth oauth\nopenclaw mcp login chukonu-web-search     # 弹浏览器完成 OAuth，token 由 host 保管\n```\n\nClaude Code：\n\n```bash\nclaude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ \\\n  --scope user --transport http\n# 然后在会话中：/mcp → chukonu-web-search → Authenticate（浏览器 OAuth）\n```\n\n> 关键：**不要**为该服务器配置 `Authorization` 头或 `CHUKONU_SEARCH_API_TOKEN`。一旦设置了静态 Authorization 头，host 会禁用 OAuth 回退，导致 `401 invalid_token`。鉴权完全交给 OAuth 流程。\n\n接入并授权后会出现两个只读 tool：`search` 和 `get_pdf_text`。\n\n## 稳定能力\n\n使用 `search` 获取网页、学术与专利证据；需要学术论文全文原文时，先让 `search` 抽取 PDF 正文，再用 `get_pdf_text` 按游标续读。`get_pdf_text` 只读取已经抽取的缓存，不会自行下载或解析 PDF。不要直接调用内部上游服务——对外入口只有 `https://search.houdutech.cn/web/mcp/`。\n\n## 何时使用本 skill\n\n- 需要外部、实时、可引用的信息（新闻、事实核查、技术细节、时效性问题）——不要凭记忆作答\n- 学术检索：论文、研究、DOI、arXiv、OpenAlex、引文、文献综述（用 `include_academic=true`）\n- 专利检索：专利、发明、申请人/权利人、发明人、IPC/CPC、公开号、FTO（用 `include_patent=true`）\n\n## 工具\n\n### `search` —— 检索入口\n\n用简洁的自然语言查询调用：\n\n```\nsearch(query=\"问题或搜索关键词\", top_k=10)\n```\n\n参数：\n\n| 参数 | 说明 |\n|---|---|\n| `query` | 自然语言查询或关键词（必填） |\n| `top_k` | 直接作答用 5-10；研究/对比类用 10-20 |\n| `include_academic` | 论文/DOI/arXiv/OpenAlex/引文/综述类任务用 `true`；学术证据会成噪声时用 `false`；否则 `null` |\n| `include_patent` | 专利/发明/申请人/发明人/IPC-CPC/公开号/FTO 类任务用 `true`；否则 `null` |\n| `rerank` | 默认 `null`（服务默认重排）；仅对排序质量要求不高的低延迟探索用 `false` |\n| `include_pdf_text` | 需要学术 PDF 正文时用 `true`；默认 `false` |\n| `pdf_text_mode` | `cached` 只读已有缓存；`sync` 允许本次搜索下载并解析 PDF；否则 `null` |\n| `pdf_max_results` | 最多为多少篇学术结果补充 PDF 正文；默认由服务决定 |\n| `pdf_max_chars_per_result` | 每篇 PDF 首次返回的正文字符上限；默认由服务决定 |\n\n示例：\n\n```\nsearch(query=\"固态电池 电解质 界面阻抗\", top_k=15, include_academic=true, include_patent=true)\n```\n\n需要论文全文时：\n\n```\nsearch(query=\"固态电池 电解质 界面阻抗\", top_k=10,\n       include_academic=true, include_pdf_text=true, pdf_text_mode=\"sync\")\n```\n\n### `get_pdf_text` —— 续读学术 PDF 正文\n\n仅在 `search(include_pdf_text=true)` 返回的 academic evidence 含有 `citation.work_id`，且 `access.next_cursor` 非空时调用：\n\n```\nget_pdf_text(work_id=\"W123\", cursor=\"<access.next_cursor>\", max_chars=8000)\n```\n\n参数：\n\n| 参数 | 说明 |\n|---|---|\n| `work_id` | OpenAlex work ID（必填）；必须从 academic evidence 的 `citation.work_id` 取得 |\n| `cursor` | 首次续读传 `access.next_cursor`，之后传上一次响应的 `next_cursor`；默认 `null` |\n| `max_chars` | 本次最多返回的字符数，默认 `8000`，服务端限制为 1–30000 |\n\n响应包含：\n\n- `status`：本次读取状态。\n- `text`：本页 PDF 正文。\n- `page_from` / `page_to` / `chunk_index`：正文定位信息。\n- `returned_chars`：本次返回的字符数。\n- `next_cursor`：下一页游标；为 `null` 时没有后续正文。\n- `partial`：是否仍有后续正文。\n- `error_code` / `error_message`：失败诊断。\n\n续读时沿用服务返回的游标，不要自行构造或修改；当 `next_cursor=null` 或现有正文已足够回答时停止。\n\n## 响应结构\n\n响应以证据为先：\n\n- `evidence[]`：按相关性排序的 `web`、`academic`、`patent` 混合证据。\n- `answerability.status`：`answerable` / `partial` / `not_answerable`。\n- `answerability.gaps[]`：明确列出缺失的证据或质量缺口。\n- `partial_failure`：任一提供方、重排或增强子任务失败时为 `true`。\n- `failures[]`：机器可读的失败详情。\n- `meta.counts`：按证据类型统计的数量。\n\n每个证据条目：\n\n- `passage.text`：可用的证据文本。\n- `citation`：label、authors、year、DOI、work_id 或 publication_number 等引用元数据。\n- `patent`：仅当 `type=\"patent\"` 时的专利结构化元数据（公开号/申请号、申请人、发明人、IPC/CPC、国家、状态、family_id、日期、patent_type、citation_count）。\n- `scores`：relevance、rank、rerank、authority、confidence 等信号。\n- `access`：开放程度、许可、OA PDF URL、PDF 状态。\n- `diagnostics`：该条证据的告警、部分失败状态、失败代码。\n\n## 决策准则\n\n- 用户给主题词/技术描述 → `search(query=…)`，按意图叠加 `include_academic` / `include_patent`\n- 需要更多召回做研究/对比 → 提高 `top_k`（10-20），而不是反复换词重搜\n- 需要论文原文且摘要不足 → `search(include_academic=true, include_pdf_text=true, pdf_text_mode=\"sync\")`\n- academic evidence 的 `access.next_cursor` 非空且仍需更多原文 → 用其 `citation.work_id` 和游标调用 `get_pdf_text`\n- 先查 `partial_failure`、`failures[]`、`answerability.gaps[]` 再组织答案\n\n## 回答规则\n\n- `answerability.status=\"not_answerable\"`：不要给自信的最终答案；说明缺什么，重新细化检索或询问用户是否继续。\n- `partial_failure=true`：使用已返回的有效证据；仅当失败来源会实质影响可信度时才提及它。\n- 学术类：优先带 DOI、venue、year、authors 与开放获取状态的 `academic` 证据；用 `citation.label`、DOI、URL 引用。\n- 专利类：优先 `patent` 证据，用 `patent.publication_number`、`patent.applicant`、`patent.inventor`、`patent.ipc_main`/`patent.cpc_main` 及公开/申请日期；不要在返回的 `patent.status` 之外推断法律状态。\n- PDF 正文：优先用 `page_from` / `page_to` 等定位信息引用；不要把摘要、搜索片段与 PDF 原文混为一谈。\n- 时效性问题：检查 `published_date`，优先较新证据；日期陈旧或缺失时说明不确定性。\n- 所有论断都要落到 `passage.text` 上；不要臆造来源、元数据、论断、引文、申请人、发明人、日期、许可或专利状态。绝不在日志/回答/示例/错误信息中泄露 token。\n\n## 错误处理\n\n| code | 应对 |\n|---|---|\n| `UNAUTHORIZED` / `401 invalid_token` | 重新 OAuth 授权：`openclaw mcp login chukonu-web-search`，或 Claude Code `/mcp → chukonu-web-search → Authenticate`。检查有没有误配静态 Authorization 头 |\n| `QUOTA_EXCEEDED` | 已达配额；告知用户重置时间或稍后重试 |\n| `INVALID_ARGUMENT` | 核对参数（`top_k` 范围、`include_academic`/`include_patent` 取值）后重试 |\n| `BACKEND_UNAVAILABLE` | 上游暂时不可用，稍后重试 |\n| `WORK_ID_MISSING` | 必须从 academic evidence 的 `citation.work_id` 传入有效 work ID |\n| `PDF_TEXT_TIMEOUT` | PDF 正文读取超时；稍后重试或基于已经返回的证据作答 |\n| `PDF_TEXT_READ_FAILED` | 检查 `work_id`、`cursor` 和 PDF 抽取状态；必要时重新用 `search(include_pdf_text=true, pdf_text_mode=\"sync\")` 初始化正文缓存 |\n\nFile v1.0.2:_meta.json\n\n{\n  \"ownerId\": \"kn7dq96kwex3nsqjg8bnbnz04h8a39jr\",\n  \"slug\": \"chukonu-web-search\",\n  \"version\": \"1.0.2\",\n  \"publishedAt\": 1784194258408\n}\n\nFile v1.0.2:skill-card.md\n\n## Description: <br>\nChukonu Web Search connects an agent to a remote OAuth-protected MCP server for read-only web, academic, and patent search with structured evidence and diagnostics. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[hhhwor](https://clawhub.ai/user/hhhwor) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and agent users use this skill to configure an OAuth MCP search integration and guide agents through evidence-based web, academic, patent, and PDF-text retrieval workflows. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: Search queries and PDF lookup requests are sent to a remote third-party service. <br>\nMitigation: Use the host-managed OAuth flow, review query sensitivity before use, and avoid static Authorization headers or tokens. <br>\nRisk: Search providers, reranking, or PDF extraction can partially fail or return incomplete evidence. <br>\nMitigation: Check answerability status, gaps, partial_failure, failures, and diagnostics before relying on results. <br>\n\n\n## Reference(s): <br>\n- [Chukonu Web Search on ClawHub](https://clawhub.ai/hhhwor/skills/chukonu-web-search) <br>\n- [Publisher Profile](https://clawhub.ai/user/hhhwor) <br>\n- [Chukonu Web Search MCP Endpoint](https://search.houdutech.cn/web/mcp/) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [text, markdown, shell commands, configuration, guidance] <br>\n**Output Format:** [Markdown guidance with inline shell commands, MCP tool parameters, and structured response fields] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [Search outputs may include evidence arrays, answerability status, gaps, partial-failure diagnostics, citations, patent metadata, and PDF text cursors.] <br>\n\n## Skill Version(s): <br>\n1.0.2 (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 v1.0.1: 3 files, 5015 bytes\n\nFiles: skill-card.md (2411b), SKILL.md (6955b), _meta.json (137b)\n\nFile v1.0.1:SKILL.md\n\n---\nname: chukonu-web-search\nversion: 0.2.0\ndescription: \"网络搜索：通过 Chukonu remote MCP server 获取面向 Agent 的网页、学术与专利检索证据。自然语言查询，返回证据优先的结构化结果（evidence[]、answerability 可答性、partial_failure 部分失败诊断），可选学术/专利意图开关与重排。只读、OAuth 登录。\"\ntags: [web, search, research, academic, patent, mcp, evidence, oauth]\nmetadata:\n  transport: streamable-http\n  requires:\n    mcp:\n      - name: chukonu-web-search\n        url: \"https://search.houdutech.cn/web/mcp/\"\n        transport: streamable-http\n        auth: oauth\n  setup:\n    # OpenClaw\n    - \"openclaw mcp add chukonu-web-search --url https://search.houdutech.cn/web/mcp/ --transport streamable-http --auth oauth\"\n    - \"openclaw mcp login chukonu-web-search\"\n    # Claude Code（本次实践；切勿配置静态 Authorization 头，否则会禁用 OAuth）\n    - \"claude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ --scope user --transport http\"\n    - \"在 Claude Code 中：/mcp → 选择 chukonu-web-search → Authenticate 完成浏览器 OAuth\"\n---\n\n# chukonu-web-search\n\n通过 Chukonu 的 **remote MCP server** 检索网页、学术与专利证据。本 skill 不调用本地二进制，也**不需要 API key / 静态 token** —— 它把 host（OpenClaw 或 Claude Code）接到托管的 `chukonu-web-search` MCP server，**OAuth 登录、只读检索**。\n\n## 一次性接入（首次使用前）\n\n端点：`https://search.houdutech.cn/web/mcp/`（streamable-http）。该端点通过标准 OAuth 2.0 保护：动态客户端注册 + 授权码 + PKCE(S256) + refresh_token，scope `search:read`。host 会自动发现并注册，无需手工填 client_id/secret。\n\nOpenClaw：\n\n```bash\nopenclaw mcp add chukonu-web-search \\\n  --url https://search.houdutech.cn/web/mcp/ \\\n  --transport streamable-http \\\n  --auth oauth\nopenclaw mcp login chukonu-web-search     # 弹浏览器完成 OAuth，token 由 host 保管\n```\n\nClaude Code：\n\n```bash\nclaude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ \\\n  --scope user --transport http\n# 然后在会话中：/mcp → chukonu-web-search → Authenticate（浏览器 OAuth）\n```\n\n> 关键：**不要**为该服务器配置 `Authorization` 头或 `CHUKONU_SEARCH_API_TOKEN`。一旦设置了静态 Authorization 头，host 会禁用 OAuth 回退，导致 `401 invalid_token`。鉴权完全交给 OAuth 流程。\n\n接入并授权后会出现只读 tool：`search`（稳定能力）。\n\n## 稳定能力\n\n仅使用稳定的 `search` 能力。PDF 全文抽取与续读不在本版本；请基于常规搜索证据作答：网页摘要/正文、学术摘要与元数据、专利摘要与元数据。不要直接调用内部上游服务——对外入口只有 `https://search.houdutech.cn/web/mcp/`。\n\n## 何时使用本 skill\n\n- 需要外部、实时、可引用的信息（新闻、事实核查、技术细节、时效性问题）——不要凭记忆作答\n- 学术检索：论文、研究、DOI、arXiv、OpenAlex、引文、文献综述（用 `include_academic=true`）\n- 专利检索：专利、发明、申请人/权利人、发明人、IPC/CPC、公开号、FTO（用 `include_patent=true`）\n\n## 工具\n\n### `search` —— 唯一检索入口\n\n用简洁的自然语言查询调用：\n\n```\nsearch(query=\"问题或搜索关键词\", top_k=10)\n```\n\n参数：\n\n| 参数 | 说明 |\n|---|---|\n| `query` | 自然语言查询或关键词（必填） |\n| `top_k` | 直接作答用 5-10；研究/对比类用 10-20 |\n| `include_academic` | 论文/DOI/arXiv/OpenAlex/引文/综述类任务用 `true`；学术证据会成噪声时用 `false`；否则 `null` |\n| `include_patent` | 专利/发明/申请人/发明人/IPC-CPC/公开号/FTO 类任务用 `true`；否则 `null` |\n| `rerank` | 默认 `null`（服务默认重排）；仅对排序质量要求不高的低延迟探索用 `false` |\n\n示例：\n\n```\nsearch(query=\"固态电池 电解质 界面阻抗\", top_k=15, include_academic=true, include_patent=true)\n```\n\n## 响应结构\n\n响应以证据为先：\n\n- `evidence[]`：按相关性排序的 `web`、`academic`、`patent` 混合证据。\n- `answerability.status`：`answerable` / `partial` / `not_answerable`。\n- `answerability.gaps[]`：明确列出缺失的证据或质量缺口。\n- `partial_failure`：任一提供方、重排或增强子任务失败时为 `true`。\n- `failures[]`：机器可读的失败详情。\n- `meta.counts`：按证据类型统计的数量。\n\n每个证据条目：\n\n- `passage.text`：可用的证据文本。\n- `citation`：label、authors、year、DOI、work_id 或 publication_number 等引用元数据。\n- `patent`：仅当 `type=\"patent\"` 时的专利结构化元数据（公开号/申请号、申请人、发明人、IPC/CPC、国家、状态、family_id、日期、patent_type、citation_count）。\n- `scores`：relevance、rank、rerank、authority、confidence 等信号。\n- `access`：开放程度、许可、OA PDF URL、PDF 状态。\n- `diagnostics`：该条证据的告警、部分失败状态、失败代码。\n\n## 决策准则\n\n- 用户给主题词/技术描述 → `search(query=…)`，按意图叠加 `include_academic` / `include_patent`\n- 需要更多召回做研究/对比 → 提高 `top_k`（10-20），而不是反复换词重搜\n- 先查 `partial_failure`、`failures[]`、`answerability.gaps[]` 再组织答案\n\n## 回答规则\n\n- `answerability.status=\"not_answerable\"`：不要给自信的最终答案；说明缺什么，重新细化检索或询问用户是否继续。\n- `partial_failure=true`：使用已返回的有效证据；仅当失败来源会实质影响可信度时才提及它。\n- 学术类：优先带 DOI、venue、year、authors 与开放获取状态的 `academic` 证据；用 `citation.label`、DOI、URL 引用。\n- 专利类：优先 `patent` 证据，用 `patent.publication_number`、`patent.applicant`、`patent.inventor`、`patent.ipc_main`/`patent.cpc_main` 及公开/申请日期；不要在返回的 `patent.status` 之外推断法律状态。\n- 时效性问题：检查 `published_date`，优先较新证据；日期陈旧或缺失时说明不确定性。\n- 所有论断都要落到 `passage.text` 上；不要臆造来源、元数据、论断、引文、申请人、发明人、日期、许可或专利状态。绝不在日志/回答/示例/错误信息中泄露 token。\n\n## 错误处理\n\n| code | 应对 |\n|---|---|\n| `UNAUTHORIZED` / `401 invalid_token` | 重新 OAuth 授权：`openclaw mcp login chukonu-web-search`，或 Claude Code `/mcp → chukonu-web-search → Authenticate`。检查有没有误配静态 Authorization 头 |\n| `QUOTA_EXCEEDED` | 已达配额；告知用户重置时间或稍后重试 |\n| `INVALID_ARGUMENT` | 核对参数（`top_k` 范围、`include_academic`/`include_patent` 取值）后重试 |\n| `BACKEND_UNAVAILABLE` | 上游暂时不可用，稍后重试 |\n\nFile v1.0.1:_meta.json\n\n{\n  \"ownerId\": \"kn7dq96kwex3nsqjg8bnbnz04h8a39jr\",\n  \"slug\": \"chukonu-web-search\",\n  \"version\": \"1.0.1\",\n  \"publishedAt\": 1783470917562\n}\n\nFile v1.0.1:skill-card.md\n\n## Description: <br>\nChukonu Web Search connects an agent to the Chukonu remote MCP service for read-only web, academic, and patent search with evidence-first results. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[hhhwor](https://clawhub.ai/user/hhhwor) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and agents use this skill to perform current web, academic, and patent searches through an OAuth-protected remote MCP server and answer from returned evidence. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: Search queries and returned results pass through a third-party Chukonu/Houdutech remote MCP service. <br>\nMitigation: Install only when that data flow is acceptable, avoid sensitive queries unless appropriate, and remove the MCP server or revoke OAuth access when it is no longer needed. <br>\nRisk: Static Authorization headers or token variables can interfere with the OAuth flow and expose credentials unnecessarily. <br>\nMitigation: Use host-managed OAuth login, avoid static Authorization headers, and keep OAuth credentials managed by OpenClaw or Claude Code. <br>\nRisk: Search results may be partial, unavailable, stale, or insufficient for a confident answer. <br>\nMitigation: Check answerability, gaps, partial failure diagnostics, and evidence dates before presenting conclusions. <br>\n\n\n## Reference(s): <br>\n- [ClawHub skill page](https://clawhub.ai/hhhwor/skills/chukonu-web-search) <br>\n- [Chukonu MCP endpoint](https://search.houdutech.cn/web/mcp/) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Text, Markdown, Shell commands, Configuration, Guidance] <br>\n**Output Format:** [Markdown guidance with MCP setup commands and structured search evidence] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [Search results may include evidence, answerability status, partial failure diagnostics, citations, and web, academic, or patent metadata.] <br>\n\n## Skill Version(s): <br>\n1.0.1 (source: server release metadata; artifact frontmatter reports 0.2.0) <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: 3 files, 3548 bytes\n\nFiles: skill-card.md (2056b), SKILL.md (4994b), _meta.json (137b)\n\nFile v1.0.0:SKILL.md\n\n---\nname: chukonu-web-search\ndescription: Use the Chukonu Web Search service at https://search.houdutech.cn/web/ for agent-ready web, academic, and patent search evidence. Trigger when a task needs external, current, citable, academic, or patent information; return and reason from structured evidence, answerability gaps, and partial-failure diagnostics.\nversion: 0.1.0\nmetadata:\n  openclaw:\n    primaryEnv: CHUKONU_SEARCH_API_TOKEN\n    envVars:\n      - name: CHUKONU_SEARCH_MCP_URL\n        required: false\n        description: Optional MCP endpoint, https://search.houdutech.cn/web/mcp/.\n      - name: CHUKONU_SEARCH_BASE_URL\n        required: false\n        description: Optional REST base URL, https://search.houdutech.cn/web/.\n      - name: CHUKONU_SEARCH_API_TOKEN\n        required: false\n        description: Optional bearer token for the /mcp and /search endpoints. Never print or expose this token.\n---\n\n# Chukonu Web Search\n\nUse this skill to search the Chukonu Web Search service and give the user evidence-backed answers. Prefer the configured MCP server when available; use the REST `/search` endpoint only as a fallback.\n\n## Stable Capability\n\nUse only the stable `search` capability.\n\nPDF full-text extraction and continuation are not part of this skill version. Answer from the normal search evidence: web snippets/content, academic abstracts and metadata, and patent abstracts and metadata.\n\n## Connection\n\nUse one of these connection paths:\n\n- MCP: `CHUKONU_SEARCH_MCP_URL`, `https://search.houdutech.cn/web/mcp/`.\n- REST fallback: `CHUKONU_SEARCH_BASE_URL` is `https://search.houdutech.cn/web/`, then `POST https://search.houdutech.cn/web/search`.\n\nIf the service requires auth, send `Authorization: Bearer $CHUKONU_SEARCH_API_TOKEN`. Never reveal the token in logs, answers, examples, or error messages.\n\nDo not call internal upstream services directly. The public product surface is `https://search.houdutech.cn/web/`; do not query internal-only hosts or ports.\n\n## Search Request\n\nCall `search` with a concise natural-language query:\n\n```json\n{\n  \"query\": \"question or search terms\",\n  \"top_k\": 10,\n  \"include_academic\": null,\n  \"include_patent\": null,\n  \"rerank\": null\n}\n```\n\nParameter guidance:\n\n- `include_academic`: use `true` for paper, research, DOI, arXiv, OpenAlex, citation, or literature-review tasks; use `false` when academic evidence would be noise; otherwise leave `null`.\n- `include_patent`: use `true` for patent, invention, applicant, assignee, inventor, IPC/CPC, publication number, or freedom-to-operate style tasks; otherwise leave `null`.\n- `rerank`: leave `null` for service default. Use `false` only for low-latency exploration where ranking quality matters less.\n- `top_k`: use 5-10 for direct answers, 10-20 for research or comparison tasks.\n\n## Response Contract\n\nThe response is evidence-first:\n\n- `evidence[]`: mixed `web`, `academic`, and `patent` evidence sorted by relevance.\n- `answerability.status`: `answerable`, `partial`, or `not_answerable`.\n- `answerability.gaps[]`: explicit missing evidence or quality gaps.\n- `partial_failure`: true when any provider, rerank, or enrichment subtask failed.\n- `failures[]`: machine-readable failure details.\n- `meta.counts`: counts by evidence type.\n\nEach evidence item has:\n\n- `passage.text`: the usable evidence text.\n- `citation`: citation-oriented metadata such as label, authors, year, DOI, work_id, or publication_number.\n- `patent`: patent-only structured metadata when `type=\"patent\"`, including publication/application numbers, applicant, inventor, IPC/CPC, country, status, family_id, dates, patent_type, and citation_count.\n- `scores`: relevance, rank, rerank, authority, and confidence signals.\n- `access`: openness, license, OA PDF URL, and PDF status.\n- `diagnostics`: warnings, partial state, and failure code for that evidence item.\n\n## Answering Rules\n\nFirst inspect `partial_failure`, `failures[]`, and `answerability.gaps[]`.\n\nIf `answerability.status=\"not_answerable\"`, do not present a confident final answer. Say what is missing and either run a refined search or ask the user whether to continue.\n\nIf `partial_failure=true`, use any valid evidence that was returned, but mention the failed source only when it materially affects confidence.\n\nFor academic answers, prefer `academic` evidence with DOI, venue, year, authors, and open-access status. Cite with `citation.label`, DOI, and URL when present.\n\nFor patent answers, prefer `patent` evidence and use `patent.publication_number`, `patent.applicant`, `patent.inventor`, `patent.ipc_main` or `patent.cpc_main`, and publication/application dates. Do not infer legal status beyond the returned `patent.status`.\n\nFor time-sensitive questions, check `published_date` and favor recent evidence. If dates are stale or absent, state the uncertainty.\n\nGround claims in `passage.text`. Do not invent sources, metadata, claims, citations, applicants, inventors, dates, licenses, or patent status that are not present in the evidence.\n\nFile v1.0.0:_meta.json\n\n{\n  \"ownerId\": \"kn7dq96kwex3nsqjg8bnbnz04h8a39jr\",\n  \"slug\": \"chukonu-web-search\",\n  \"version\": \"1.0.0\",\n  \"publishedAt\": 1783456293356\n}\n\nFile v1.0.0:skill-card.md\n\n## Description: <br>\nUses the Chukonu Web Search service to retrieve agent-ready web, academic, and patent search evidence with answerability gaps and partial-failure diagnostics. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[hhhwor](https://clawhub.ai/user/hhhwor) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers, researchers, and external agents use this skill when they need current, citable web, academic, or patent evidence and need to reason from returned gaps, failures, and structured citations. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: Search queries and any configured bearer token are sent to the external Chukonu search service. <br>\nMitigation: Use this skill only for tasks where that external data flow is acceptable, avoid private or sensitive queries unless the service is trusted, and never expose configured bearer tokens in logs or answers. <br>\n\n\n## Reference(s): <br>\n- [Chukonu Web Search](https://search.houdutech.cn/web/) <br>\n- [Chukonu Web Search MCP endpoint](https://search.houdutech.cn/web/mcp/) <br>\n- [ClawHub skill page](https://clawhub.ai/hhhwor/skills/chukonu-web-search) <br>\n- [Publisher profile](https://clawhub.ai/user/hhhwor) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Text, Markdown, API Calls, Guidance] <br>\n**Output Format:** [Markdown answers grounded in structured JSON search evidence] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [Search evidence can include evidence[], answerability status and gaps, partial failure diagnostics, citations, academic metadata, and patent metadata.] <br>\n\n## Skill Version(s): <br>\n1.0.0 (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>","readmeExcerpt":"Skill: chukonu-web-search Owner: hhhwor Summary: 通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利、金融和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及财经新闻与金融研究、法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。 Tags: latest:1.1.4 Version history: v1.1.4 | 2026-09-09T01:02:26.993Z | user chukonu-web-search 1.1.4 - 新增财经新闻与金融文档检索及研究支持，source_types 增加 financial，支持金融新闻、结构化行情、财报等能力。 - 文档指引","codeSnippets":[],"executableExamples":[{"language":"bash","snippet":"openclaw mcp add chukonu-web-search \\\n  --url https://search.houdutech.cn/web/mcp/ \\\n  --transport streamable-http \\\n  --auth oauth\nopenclaw mcp login chukonu-web-search"},{"language":"bash","snippet":"claude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ \\\n  --scope user --transport http\n# 然后在会话中：/mcp → chukonu-web-search → Authenticate"},{"language":"bash","snippet":"codex mcp add chukonu-web-search --url https://search.houdutech.cn/web/mcp/\ncodex mcp login chukonu-web-search\ncodex mcp list"},{"language":"json","snippet":"{\n  \"query\": \"固态电池硫化物电解质近五年的关键路线\"\n}"},{"language":"json","snippet":"{\n  \"query\": \"《中华人民共和国民法典》第1084条\",\n  \"limit\": 10,\n  \"source_types\": [\"legal\"],\n  \"filters\": {\n    \"legal_status\": \"现行有效\",\n    \"jurisdictions\": [\"CN\"]\n  }\n}"},{"language":"json","snippet":"{\n  \"query\": \"劳动合同解除的法定情形和近期解读\",\n  \"source_types\": [\"legal\", \"web\"],\n  \"filters\": {\n    \"legal_status\": \"现行有效\",\n    \"languages\": [\"zh\"]\n  }\n}"}],"parameters":null,"dependencies":[],"permissions":[],"extractedFiles":[{"path":"SKILL.md","content":"---\nname: chukonu-web-search\ndescription: \"通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利、金融和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及财经新闻与金融研究、法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。\"\n---\n\n# Chukonu Search + Research\n\n通过 Chukonu remote MCP 调用公开的 `search` 和 `research` 工具。无需本地二进制、API key 或静态 token；不要调用内部工具、上游服务或内部端口。法律法规检索是服务内部接入的 FY MCP provider，客户端不要单独配置它、传入其凭据或尝试调用其工具。\n\n## 一次性接入\n\n使用 `https://search.houdutech.cn/web/mcp/` streamable-http 端点。该端点通过标准 OAuth 2.0 保护，采用动态客户端注册、授权码、PKCE(S256) 和 refresh token，scope 为 `search:read`。让 MCP host 自动发现并注册，不要手工填写 client ID 或 client secret。\n\nOpenClaw：\n\n```bash\nopenclaw mcp add chukonu-web-search \\\n  --url https://search.houdutech.cn/web/mcp/ \\\n  --transport streamable-http \\\n  --auth oauth\nopenclaw mcp login chukonu-web-search\n```\n\nClaude Code：\n\n```bash\nclaude mcp add chukonu-web-search https://search.houdutech.cn/web/mcp/ \\\n  --scope user --transport http\n# 然后在会话中：/mcp → chukonu-web-search → Authenticate\n```\n\nCodex：\n\n```bash\ncodex mcp add chukonu-web-search --url https://search.houdutech.cn/web/mcp/\ncodex mcp login chukonu-web-search\ncodex mcp list\n```\n\n在 Codex TUI 中用 `/mcp` 查看连接状态。若 OAuth 浏览器与 Codex 不在同一主机，先为 Codex 配置可回调的 `mcp_oauth_callback_url`，再登录；不要输出或持久化授权码、回调 URL、access token 或 refresh token。\n\n不要为该服务器配置静态 `Authorization` 头或 `CHUKONU_SEARCH_API_TOKEN`。静态 Authorization 头会使 host 禁用 OAuth 回退，并可能导致 `401 invalid_token`。把 OAuth token 完全交给 host 保管，不要在日志、回答、示例或错误信息中泄露。\n\n## 选择工具与来源\n\n- 用 `search` 处理秒级发现、法规定位和获取后续研究所需的 `research_seed.search_id`。\n- 用 `research` 处理关键主张核验、反证、跨来源冲突、全文定位或证据覆盖评估。必须先 `search`，再用返回的 `search_id` 启动任务。\n- 本技能使用的 `source_types` 为 `web`、`academic`、`patent`、`legal` 和 `financial`。省略时服务自动路由；财经新闻归入 `financial`，法律意图可在法律 provider 启用时与通用网页并行召回。\n- 用 `source_types: [\"legal\"]` 只检索中国法律法规；用 `source_types: [\"legal\", \"web\"]` 同时获取法条和通用网页背景。\n- 用 `source_types: [\"financial\"]` 检索财经新闻和金融文档；宁夏服务已接入 `nws_news` 新闻库，其他金融来源取决于服务端配置。仅查金融时显式指定来源；需要网页原文或背景时用 `[\"financial\", \"web\"]`。\n- 金融问题需要进一步研究时，用 `research` 的 `profile: \"financial_research\"`；结构化股票数据是否可用要检查返回的 `financial_plan`，不能仅凭 profile 已接受就认定已启用行情查询。\n- 对法律问题尽量写明法名和条号，例如 `《中华人民共和国民法典》第1084条`；仅把原始法条证据用于法律效力或条文结论。\n- 不要把搜索相关性分数解释为事实置信度，也不要把检索结果或本技能当作法律意见。\n\n## `search`\n\n### 请求\n\n最小调用：\n\n```json\n{\n  \"query\": \"固态电池硫化物电解质近五年的关键路线\"\n}\n```\n\n只检索现行有效的中国法规：\n\n```json\n{\n  \"query\": \"《中华人民共和国民法典》第1084条\",\n  \"limit\": 10,\n  \"source_types\": [\"legal\"],\n  \"filters\": {\n    \"legal_status\": \"现行有效\",\n    \"jurisdictions\": [\"CN\"]\n  }\n}\n```\n\n混合检索法条和网页解读：\n\n```json\n{\n  \"query\": \"劳动合同解除的法定情形和近期解读\",\n  \"source_types\": [\"legal\", \"web\"],\n  \"filters\": {\n    \"legal_status\": \"现行有效\",\n    \"languages\": [\"zh\"]\n  }\n}\n```\n\n只检索财经新闻和金融文档：\n\n```json\n{\n  \"query\": \"贵州茅台 分红\",\n  \"source_types\": [\"financial\"],\n  \"filters\": {\n    \"published_from\": \"2026-08-01\",\n    \"published_to\": \"2026-08-31\"\n  }\n}\n```\n\n遵守以下边界：\n\n- `query` 必填；保持简洁、具体。`limit` 是最终全局返回数，范围为 1–20。\n- 本技能使用的 `filters` 包括 `published_from`、`published_to`、`languages`、`jurisdictions`、`legal_status` 和 `patent`。`lega"},{"path":"_meta.json","content":"{\n  \"ownerId\": \"kn7dq96kwex3nsqjg8bnbnz04h8a39jr\",\n  \"slug\": \"chukonu-web-search\",\n  \"version\": \"1.1.4\",\n  \"publishedAt\": 1788915746993\n}"},{"path":"skill-card.md","content":"## Description:\n\nUses the Chukonu remote MCP search and research tools to retrieve web, academic, patent, financial, and Chinese legal evidence for cited real-time research and verification tasks.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[hhhwor](https://clawhub.ai/user/hhhwor)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers and agents use this skill when they need externally sourced, current, citable evidence for search, fact checking, counter-evidence, Chinese legal retrieval, financial-news research, PDF review, and persistent multi-turn research workflows.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: Search queries and research objectives are sent to the Chukonu remote MCP service.\n\nMitigation: Avoid submitting sensitive information unless the user accepts that remote processing; use the normal OAuth flow and let the host manage tokens.\n\nRisk: Static Authorization headers, API tokens, or leaked OAuth material could create authentication and credential-handling risk.\n\nMitigation: Do not configure static Authorization headers or CHUKONU_SEARCH_API_TOKEN, and do not print or persist authorization codes, callback URLs, access tokens, or refresh tokens.\n\nRisk: Legal and financial results may have incomplete coverage, unsupported filters, missing public links, or source-specific limitations.\n\nMitigation: Check retrieval assessments, failures, filter execution, legal status, and financial data plans before making claims; present limitations and use primary or professional review for consequential decisions.\n\n## Reference(s):\n\n- [ClawHub skill page](https://clawhub.ai/hhhwor/skills/chukonu-web-search)\n- [Chukonu remote MCP endpoint](https://search.houdutech.cn/web/mcp/)\n\n## Skill Output:\n\n**Output Type(s):** [Text, Markdown, Shell commands, Configuration, Guidance]\n\n**Output Format:** [Markdown responses with citations, inline JSON examples, and shell command examples]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May include cited evidence, retrieval limitations, research status, and OAuth setup guidance.]\n\n## Skill Version(s):\n\n1.1.4 (source: server release evidence)\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."}],"languages":[],"docsSourceLabel":"CLAWHUB","editorialOverview":"通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利、金融和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及财经新闻与金融研究、法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。 Skill: chukonu-web-search Owner: hhhwor Summary: 通过 Chukonu remote MCP 的 search 与 research 获取网页、学术、专利、金融和中国法律法规证据。用于需要外部、实时、可引用信息的搜索任务，以及财经新闻与金融研究、法律法规检索、事实核验、反证检索、PDF 全文深读、覆盖评估或持久化多轮研究；依据结构化 evidence、retrieval_assessment 与 research dossier 作答。 Tags: latest:1.1.4 Version history: v1.1.4 | 2026-09-09T01:02:26.993Z | user chukonu-web-search 1.1.4 - 新增财经新闻与金融文档检索及研究支持，source_types 增加 financial，支持金融新闻、结构化行情、财报等能力。 - 文档指引","editorialQuality":{"score":100,"threshold":65,"status":"ready","wordCount":1153,"uniquenessScore":50,"reasons":[]}},"media":{"evidence":{"source":"no-media","verified":false,"confidence":"low","updatedAt":"2026-10-11T11:29:35.766Z","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":"2026-10-11T11:29:35.766Z","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":"2026-10-11T14:15:38.342Z","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"}]}}}