remember

Remember

Recall past work from Claude Code session history. The user is trying to recover context — a decision, a fix, a paper, an approach — not run a database query.

What can be remembered

Hooks determine what's in the searchable index:

• Research activity — every URL fetched, every web search query
• Session lifecycle — compactions (with git state snapshot), session ends, agent completions
• Code changes — file edits, bash commands, git commits with type classification (when CARTOGRAPHER_LOG_TOOL_USE=true)
• Raw transcripts — full conversation text (slower, searched as fallback)

Search facets

Results include faceted summaries (project, source, event type, time range). Use these to narrow searches:

• --project <name> — filter to a specific project
• Commit types — git commits are classified as feature, fix, refactor, enhancement, docs, test, chore, perf, ci, style, revert, or other. These appear in summaries as [feature] Commit abc1234: ... and are searchable as keywords.
• Event types in results: git_commit, research_fetch, research_search, milestone_session_end_*, milestone_compaction_*, tool_file_edit, tool_bash
• Session end events now include git branch, dirty file count, and session event count — useful for finding "what was I working on last time"

IMPORTANT: Use the search script

Do NOT freestyle grep or jq commands. Always use the unified search script.

Step 1: Run the search

Think about what the user is trying to recall, then translate to search terms.

bash ~/Documents/dev/session-cartographer/scripts/cartographer-search.sh "<search terms>"

If the user mentioned a specific project, add --project <name>:

bash ~/Documents/dev/session-cartographer/scripts/cartographer-search.sh "<terms>" --project scrutinizer

For more results, add --limit 25 or --limit 50. If the user says "more" or "keep going" after seeing results, re-run with a higher limit:

bash ~/Documents/dev/session-cartographer/scripts/cartographer-search.sh "<same terms>" --limit 30

Wildcard prefix search works: shader* matches shader, shaders, shaderlab, etc.

Step 2: Present results

Show results as-is from the script output. Keep it scannable.

Step 3: Read the transcript when needed

Search results are summaries. When you need full context, read the transcript file directly — don't wait for the user to ask.

Find the transcript path — it's in the search result as transcript:. If missing, resolve from session::

find ~/.claude/projects -name "<session-id>.jsonl" 2>/dev/null

Then read around the relevant moment:

jq -c 'select(.type == "user" or .type == "assistant") | select(.message.content | type == "string") | {type, timestamp, content: .message.content[:500]}' <transcript_path> | grep -A5 -B5 "<keyword>"

Or jump to a specific message by UUID:

jq 'select(.uuid == "<uuid>" or .parentUuid == "<uuid>")' <transcript_path>

The search result is the map. The transcript is the territory.

Examples

/remember that paper about foveated rendering
/remember what we decided about the shader approach
/remember the commit that fixed blur
/remember Blauch collaboration notes
/remember recent feature commits --project scrutinizer
/remember what was I working on last session