caixu-ingest-materials

Search Personal Asset Library. Use when the user wants to search or filter an existing 材序 asset library at the query stage, including “查我有哪些材料”“看哪些可复用”“按类型筛资产”“用自然语言找相关材料”. Prefer caixu-skill when the user asks for the full end-to-end mainline or is unsure which stage to run. This skill normalizes filters into precise retrieval inputs, uses agent tags plus FTS by default, and only calls the optional semantic vector retrieval function when the user explicitly asks for similar or related materials.

Maintain Personal Asset Library. Use when the user wants to inspect, review, patch, archive, restore, or verify a personal material asset library after initial build, including “看库概览”“修一条资产”“归档这条不该进库的材料”“看待复核队列”. Prefer caixu-skill when the user asks for the full end-to-end mainline or is unsure which stage to run. This skill loads library overviews and review queues from caixu-data-mcp, applies minimal maintenance actions such as patching or archiving one asset at a time, and re-checks query results. Do not use it for fresh ingest, asset extraction, lifecycle judgment, package building, or submission.

Build Personal Asset Library. Use when the user wants to convert parsed materials into asset cards and a queryable library at the asset-build stage, including “建资产库”“生成资产卡”“去重整理版本”. Prefer caixu-skill when the user asks for the full end-to-end mainline or is unsure which stage to run. This skill creates a build_asset_library pipeline run, reads parsed files from caixu-data-mcp, extracts canonical asset cards from trustworthy parsed text, merges likely duplicate versions conservatively, persists assets and merged groups, and completes the pipeline run with library_id. Do not use it for querying, lifecycle judgment, package creation, or submission.

Asset Library Skill. Use when the user expresses the overall end-to-end intent in one request, including “把这些材料建成资产库”“列出未来 60 天需要续办或补办的事项”“生成暑期实习申请材料包”, when the user wants to search materials by natural language but is unsure whether the local retrieval environment is ready, when the user is unsure which phase skill to run, or when the environment may not be installed yet. This skill routes only to the current child skill, sends first-time installation agents to references/install.md when keys, retrieval config, or runtime setup are missing, explains stage boundaries and next steps, and does not directly call MCP tools, extract files, build packages, or submit forms.

USE when implementing data ingestion for Agnost AI analytics. Contains API reference, SDK guides for Python and TypeScript, and code examples for tracking AI conversations, MCP server events, and user interactions.

Investigative intelligence — document search, entity extraction, and relationship graphing. Analyze document corpuses to find connections between people, organizations, and identifiers.

中小学教学文档智能生成工具。当用户需要制作课件（PPT）、编写教学设计（教案）、或创建学生任务单（导学案）时使用此技能。支持数学、语文等学科的备课、课堂巩固、课后作业等教学场景的文档生成。

Auto-analyze URLs/info for OpenClaw strategic value, classify, create Obsidian notes, update memory. Use when user shares a URL, article, tweet, or any external info source.

Check Document Renewal Requirements. Use when the user wants a lifecycle diagnosis for an existing asset library at the lifecycle stage, including “看未来 60 天续办”“查缺件”“判断能不能提交某个目标场景”. Prefer caixu-skill when the user asks for the full end-to-end mainline or is unsure which stage to run. This skill loads assets and a versioned RuleProfileBundle, asks an agent to produce a complete CheckLifecycleData decision, validates readiness and asset references with shared rules helpers, records an audit sidecar, and persists the lifecycle run only when the decision is structurally valid.

Build Application Document Package. Use when the user wants export files and a submission bundle from an existing library at the packaging stage, including “生成材料包”“导出资产总表和续办清单”“准备提交包”. Prefer caixu-skill when the user asks for the full end-to-end mainline or is unsure which stage to run. This skill loads the latest validated lifecycle result, a RuleProfileBundle, assets and parsed files, asks an agent to choose package contents, validates the decision, runs deterministic docgen exports, persists the package run with an audit sidecar, and prepares the final bundle for downstream use.

name: article-factory-wechat

Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, negative parallelisms, and excessive conjunctive phrases.