doctorlisted

# Doctor When this skill is invoked, run the helper directly and return its grouped findings or the shortest actionable failure. Resolve `PLUGIN_ROOT` from this selected skill's installed absolute source path: it is the directory two levels above this `SKILL.md`. Substitute that absolute path literally in the helper invocation below; never infer it from the project working directory or hardcode a marketplace cache version. `doctor.sh` is STRICTLY read-only: it never writes to docs, the memory store, MEMORY.md, any memory file, the capture inbox, the context store, or `AGENTS.md` — every check reads, stats, or invokes another read-only helper (never `memory-write.sh`, not even its `unlock` subcommand, which itself removes a dead lock). Run exactly **one** literal Bash segment (no `export`/`env`/assignment prefix, no chaining/piping/redirection): ``` bash "<PLUGIN_ROOT>/scripts/doctor.sh" [--store <path>] ``` Pass `--store <path>` only if the user supplied one in the user's arguments; it governs ONLY the memory-module and lock-diagnostics sections (same precedence as every other memory command: explicit target > `KNOWLEDGE_MEMORY_HOME` > canonical discovery under `.agents/memory/`). The docs, context, AGENTS.md, capability-matrix, and duplicate-plugin sections always target the repository root and never take `--store`. That is the whole argv grammar — anything else is a usage error. Exit codes: `0` clean (no `WARN`/`ERROR` finding — `INFO` findings may still be present an