$npx -y skills add chacosoldier/compabob --skill index-memoryBuild or refresh the local memory-search index over memory/ and vault/, so retrieval finds notes by relevance. Use for "/index-memory", "reindex my memory", "rebuild the search index".
| 1 | # Index Memory |
| 2 | |
| 3 | Builds the local search index that the memory-search module uses. Run it after a |
| 4 | burst of note-taking, or on a schedule. |
| 5 | |
| 6 | ## Steps |
| 7 | |
| 8 | 1. Run the indexer: |
| 9 | `python3 modules/memory-search/index.py` |
| 10 | 2. Report back: how many chunks were indexed, and which backend it used. |
| 11 | **Semantic** means search-by-meaning (Ollama with an embedding model is |
| 12 | running). **Keyword** means FTS5 ranked search, the zero-setup default. |
| 13 | 3. If it used the keyword backend and the user wants search-by-meaning, point |
| 14 | them to `docs/how-to-improve-memory.md` for the one-time Ollama setup. |
| 15 | |
| 16 | Once the index exists, the `second-brain` agent queries it automatically when |
| 17 | retrieving notes. To keep it current, re-run this skill, or schedule it (see |
| 18 | `modules/memory-search/README.md`). |