What problem does it solve? When a capability is already implemented in code but its knowledge is scattered across multiple files, AI coding agents cannot retrieve it later. This Skill consolidates those dispersed sources into structured draft and final documents inside the .Knowledge/ knowledge base, then synchronizes topics, index, and routing manifests so future sessions can find the facts. ## Core Features & Use Cases - Multi-file aggregation: Parses one or more source files (code, config, docs) into a draft (<方案名>_初稿.md) and a finalized document (<方案名>_终稿.md) under .Knowledge/stock-docs/. - Multi-module detection: When input paths span two or more functional directories, it pauses and asks the user whether to generate per-module documents or a merged one. - Sub-agent orchestration: For large inputs (5+ paths, 3000+ line files, or 10000+ total lines), it can split read-only scanning across parallel sub-agents using B-mode or C-mode strategies while the main agent retains write control. - Routing synchronization: Updates .Knowledge/topics/, .Knowledge/index.md, and manifest-routing.json via the f2s-kb-build conventions after finalization. - Use Case: After shipping a payment feature spread across src/payment/checkout.ts, config, and docs, run this Skill to produce a finalized knowledge document and register its topic so future agent sessions route to it directly. ## Quick Start Ask the agent to run f2s-kb-add on the implemented feature files, for example: add src/auth/login.ts and src/auth/session.ts into the knowledge base as the auth capability.
