| name | remember |
| description | Save information to persistent memory for cross-session recall. Stores preferences, conventions, decisions, and context. |
/remember — Persistent Memory Management
$ARGUMENTS
🔴 CRITICAL RULES
- Load memory-system skill — Read
.agent/skills/memory-system/SKILL.md first
- Never auto-delete memories — Always ask user before pruning
- Keep index under 200 lines — Warn if approaching limit
- Distill, don't copy — Save insights, not full conversations
Task
Use the memory-system skill to save information:
CONTEXT:
- User wants to remember: $ARGUMENTS
- Memory location: .agent/memory/
WORKFLOW:
1. CLASSIFY the information type: user | feedback | project | reference
2. CHECK if relevant topic file exists in .agent/memory/
3. SAVE to appropriate topic file (create if needed)
4. UPDATE .agent/memory/MEMORY.md index with one-line pointer
5. CONFIRM to user what was saved
RULES:
1. Follow memory-system/SKILL.md taxonomy
2. Keep index entries under 150 characters
3. Topic files must have frontmatter (type, created, updated)
4. Don't save information derivable from code
5. Don't save temporary debug context
Expected Output
[OK] Saved to memory
Type: [user/feedback/project/reference]
File: .agent/memory/[topic-file].md
Entry: [one-line summary of what was saved]
This will be available in future sessions.
Usage Examples
/remember I prefer using bun instead of npm
/remember Our API uses JWT with httpOnly cookies
/remember The production server is at api.example.com:8080
/remember I like concise responses with tables
Generated by Agent Bridge