majiayu000 /memorize - Save Agent Memory + Update Dashboard
One command to update all memory files and Brief.md. Replaces manually asking "update your memory."
Usage
/memorize # Save current agent's memory + update Brief.md
/memorize <agent> # Save a specific agent's memory
Available Agents
planner, builder, checker, brainstorm, gal, legal, gitdude, fetcher, orca
Instructions
When invoked, perform these steps IN ORDER for the target agent:
Step 1: Identify the Agent
Determine the agent's Memory_Logs path:
- planner:
AgenTeam/Planner/Memory_Logs/ - builder:
AgenTeam/Builder/Memory_Logs/ - checker:
AgenTeam/Checker/Memory_Logs/ - brainstorm:
AgenTeam/BrainStorm/Memory_Logs/ - gal:
AgenTeam/Gal/Memory_Logs/ - legal:
AgenTeam/Legal/Memory_Logs/ - gitdude:
AgenTeam/GitDude/Memory_Logs/ - fetcher:
Library/Fetcher/Memory_Logs/ - orca:
AgenTeam/Orca/Memory_Logs/
If no agent specified, use the current agent (the one active in this session).
Step 2: Read Current State
Read these files to understand what needs updating:
Memory_Logs/Checkpoint.md— current task statusMemory_Logs/Lessons.md— existing lessonsMemory_Logs/Preferences.md— existing preferences- Latest file in
Memory_Logs/Sessions/— recent session context Dashboard/Brief.md— current project state
Step 3: Append to Sessions/
Create or append to Memory_Logs/Sessions/Session_[YYYY-MM-DD].md:
# Session — [YYYY-MM-DD]
**Agent:** [Agent Name]
**Task:** [Brief description of what was worked on]
---
## What Happened
- [Key actions taken]
- [Decisions made]
- [Files created/modified]
## Outcome
[Result of the work]
---
IMPORTANT: If a session file for today already exists, APPEND to it — never overwrite.
Step 4: Update Lessons.md
READ first, then APPEND if something was learned this session:
## [YYYY-MM-DD] - [Brief title]
- **Context:** [what happened]
- **Lesson:** [what to remember]
- **Apply when:** [future trigger]
If nothing new was learned, skip this step.
Step 5: Update Preferences.md
READ first, then APPEND if user expressed new preferences:
## [Category]
*[YYYY-MM-DD]*
- **Preference:** [detail]
If no new preferences, skip this step.
Step 6: Update Checkpoint.md
READ first, then update:
- If task is complete: mark status as complete, add to Checkpoint History
- If task is in progress: update progress steps, current state, next action
- If no active task: ensure status shows "No active task"
Step 7: Refresh Context.md
Overwrite Memory_Logs/Context.md with a fresh snapshot:
# Context — [Agent Name]
**Last updated:** [YYYY-MM-DD]
*Quick startup snapshot. Read this first instead of scanning all memory files.*
---
## Project Summary
[2-3 sentences from Dashboard/Brief.md current focus]
## Current Task
[From Checkpoint.md — status and description, or "No active task"]
## Recent Lessons
[Last 3 entries from Lessons.md — one line each]
## Active Preferences
[Key preferences from Preferences.md]
## Team Status
See Dashboard/Brief.md for full team status.
## My Recent Sessions
[1-2 line summary of last 2-3 sessions]
Step 8: Update Dashboard/Brief.md
Only if project-wide changes happened. READ first, then:
- Recent Activity: Add row:
| [YYYY-MM-DD] | [Agent] | [what happened] | - Team Status: Update agent state if changed (e.g., "Ready" → "Working on X")
- Active Projects: Update project status/next if a project changed
- Next Steps: Add or remove items if priorities shifted
If this session was internal work (no project-wide impact), skip Brief.md.
Step 9: Report
=== MEMORIZED ===
Agent: [name]
Sessions/: [created/appended Session_YYYY-MM-DD.md]
Lessons.md: [appended X new / no change]
Preferences: [appended X new / no change]
Checkpoint: [updated / no change]
Context.md: [refreshed]
Brief.md: [updated / no change]
=================
Critical Rules
- Memory is SACRED: ALWAYS read before write, ALWAYS append (except Context.md)
- Context.md IS overwritten — it's a generated snapshot, not accumulated data
- Sessions/ gets one file per date — append if same day, new file if new day
- Never delete memory — only add to it
- Brief.md is project-level — only update for project-wide changes, not routine work
- Be honest — only log lessons actually learned, not filler
Part of AutoMates.AI memory management system