apappascs Step 1: Run the search
bun run ${CLAUDE_PLUGIN_ROOT}/lib/memory-scanner.ts search "<query>" --limit 20 --context 1
Replace <query> with the user's search term. Add filters based on request:
--type user|feedback|project|reference— filter by memory type--project FILTER— filter by project name--context N— lines of context around each match (default 1)--limit N— max results (default 20)
Step 2: Present the results
Output is NDJSON (one JSON object per line). Parse each line and group results by project.
For each project with matches:
Project: project-name
| File | Type | Line | Match |
|---|---|---|---|
| filename.md | type | N | matched text (trimmed) |
Show context lines (if available) indented below each match.
If no results: Suggest broadening the search or trying alternative keywords.
Step 3: Offer actions
For each matched memory file, the user might want to:
- Read the full file — use the Read tool on the
pathfield - Edit the memory — use the Edit tool on the
pathfield - Check health — suggest
/session-memory-audit