← ClaudeAtlas

mem-searchlisted

Search the team's shared aznex memory for this repo. Use when the user asks "did we already solve this?", "why is X built this way?", "what does the team know about this file?", or before re-deriving a decision, workaround, or dead end that a teammate's session may have already captured.
shashanknidhi/aznex · ★ 3 · AI & Automation · score 57
Install: claude install-skill shashanknidhi/aznex
# mem-search — query team memory (aznex) Aznex stores memories your teammates' coding sessions produced for this repo: decisions, extracted learnings, negative results (dead ends), and summaries. Query them through the `aznex` MCP tools before re-investigating anything a teammate may have already figured out. ## Repo fingerprint Every tool takes `repo_fingerprint`: the canonical git identity `<host>/<owner>/<name>` (e.g. `github.com/acme/widget`). Derive it from the repo you're working in: ```sh git remote get-url origin ``` Strip protocol/`git@`/`.git` and normalize to `host/owner/name`. ## Tools and when to use each | Tool | Use for | |---|---| | `search_memory` | Keyword/full-text search — "did we deal with <topic> before?" Start here. | | `get_recent_context` | The latest team memories — orientation when starting broad work. | | `get_memories_by_path` | Everything the team knows about one file (repo-relative path). Use before making significant changes to an unfamiliar file. | | `get_memory` | Full record by id (narrative, facts, anchors) — after a search hit needs detail. | | `list_sessions` | Capture timeline — who worked when; useful to scope "recent" questions. | ## Workflow 1. `search_memory` with 2–3 focused keyword variants (it's FTS — try the noun, not the sentence). If a result looks relevant, `get_memory` its id for the full narrative and anchors. 2. For file-scoped questions, `get_memories_by_path` with the repo-relative path. 3. Every memory