repo-memory
by miritor
Build a multi-layer memory system for code repositories (L0 Project Overview → L1 Module Summary → L2 File Summary → L3 Method Detail), with support for first-time full indexing and incremental updates via git diff. Trigger this skill when the user mentions "repo memory", "code memory", "project memory", "repo index", "code index", "code documentation", "repo understanding", "help me remember this project", "analyze the whole repo", "understand this codebase", "build code archive", "code knowledge base", "incremental update memory", or "refresh code docs". Even if the user doesn't explicitly say "memory", trigger this skill whenever the intent is to deeply understand and continuously track a code repository. [IMPORTANT] This skill is not only for archiving and updating — it also assists development. When the user wants to implement new features, fix bugs, refactor, review code, or perform any task that requires understanding the codebase, if the .repo-memory directory already exists, prefer reading memory docs over re-reading source code. Trigger words include but are not limited to: "help me implement", "develop this feature", "modify this", "add an API", "fix this bug", "refactor", "optimize", "review code", "write a new module", etc. Whenever the user's task involves a repo that has been indexed, trigger this skill in query mode.