mnemo-memorylisted

# mnemo Memory Use mnemo MCP tools to recover and persist project knowledge across sessions. Keep the always-active project instructions authoritative; this skill provides the detailed workflow. ## 1. Verify the project Before any memory operation: 1. Resolve the Git repository root, or use the current workspace when it is not a Git repository. 2. Read `<root>/.mnemo`. 3. Continue only when it is valid JSON with a non-empty `id`. 4. Use that `id` as `project` in every mnemo tool call. If `.mnemo` is missing or invalid, tell the user to run `mnemo init` and stop the memory workflow. If mnemo tools are unavailable, report that integration is incomplete. Never create `MEMORY.md`, write into an agent's native memory directory, or use arbitrary text files as a fallback. ### Troubleshoot integration When mnemo appears misconfigured, unavailable, or inconsistent with the project marker, prefer the read-only diagnostic command before guessing: ```bash mnemo doctor --path <root> mnemo doctor --json --path <root> ``` Use the output to explain missing `.mnemo`, PATH, global instruction, MCP, hook/plugin, or store issues. Do not repair configuration automatically unless the user explicitly asks for setup or install changes. ## 2. Recover relevant context - At session start, resume, or after compaction, call `mem_context` before significant work. - When the user asks to recall past work, call `mem_context` first, then `mem_search` with focused keywords. - Use `mem_get_observati