Commandtypescript

/memory Command

Display current memory state from Memory Copilot for transparency and debugging.

View Source

Memory Dashboard

Display current memory state from Memory Copilot for transparency and debugging.

Step 1: Retrieve Memory Data

Call these tools to gather memory state:

  1. Get current initiative:

    initiative_get()

  2. Get recent memories:

    memory_list({ limit: 10 })

  3. Get agent improvement suggestions:

    memory_list({ type: 'agent_improvement', limit: 20 })

  4. If Task Copilot is linked, get progress:

    progress_summary()

  5. Get protocol violations (if Task Copilot is linked):

    protocol_violations_get({ limit: 20 })

Step 2: Display Dashboard

Format the output as a clean, scannable dashboard:

## Memory Dashboard

**Initiative:** [name]
**Status:** [status - IN PROGRESS / COMPLETED / BLOCKED]

**Focus:** [currentFocus]
**Next:** [nextAction]

### Recent Decisions
[List last 3-5 decisions, or "None recorded"]

### Recent Lessons
[List last 3-5 lessons, or "None recorded"]

### Key Files
[List keyFiles, or "None tracked"]

### Recent Memories (Last 10)
[Table format:]
Type       | Content Preview                    | Created
---------- | ---------------------------------- | ----------
decision   | [First 50 chars...]                | 2025-01-15
lesson     | [First 50 chars...]                | 2025-01-14

### Agent Improvements
[If agent improvements exist, show summary:]
**Pending:** [count] | **Approved:** [count] | **Rejected:** [count]

[Table format for pending suggestions:]
Agent | Section           | Rationale                           | Created
----- | ----------------- | ----------------------------------- | ----------
me    | Core Behaviors    | [First 40 chars...]                 | 2025-01-15
ta    | Output format     | [First 40 chars...]                 | 2025-01-14

[If no improvements: "No agent improvements recorded"]

**Storage:** ~/.claude/memory/[workspace-id]/memory.db
**Workspace ID:** [workspace-id or "auto-generated hash"]

### Task Progress (if Task Copilot linked)
[Show output from progress_summary(), or skip section if not linked]
PRDs: [count] | Tasks: [pending/in_progress/completed] | Work Products: [count]

### Protocol Violations (if Task Copilot linked)
[If protocol violations exist, show summary:]
**Total:** [count] | **Critical:** [count] | **High:** [count] | **Medium:** [count] | **Low:** [count]

[Table format for recent violations:]
Type                     | Severity | Description                     | When
------------------------ | -------- | ------------------------------- | ----------
files_read_exceeded      | high     | Read 5 files (limit: 3)         | 2025-01-12
generic_agent_used       | critical | Used "Explore" agent            | 2025-01-12

[If no violations: "No protocol violations recorded"]

Step 3: Handle Edge Cases

No Active Initiative

If initiative_get returns null or no initiative exists:

## Memory Dashboard

**Status:** No active initiative

Use `/protocol` to start fresh work or `/continue` to resume.

**Storage:** ~/.claude/memory/[workspace-id]/memory.db

No Memories

If memory_list returns empty:

### Recent Memories
No memories stored yet.

No Agent Improvements

If memory_list({ type: 'agent_improvement' }) returns empty:

### Agent Improvements
No agent improvements recorded

Bloated Initiative Warning

If the initiative contains legacy fields (completed, inProgress, blocked, resumeInstructions):

**WARNING:** This initiative uses legacy format with bloated task lists.
Consider running `initiative_slim({ archiveDetails: true })` to reduce context usage by ~75%.

Display Notes

  • Keep output compact and scannable
  • Truncate long content previews to 50 characters
  • Show timestamps in YYYY-MM-DD format
  • Group decisions and lessons separately from other memory types
  • Highlight if Task Copilot is linked (taskCopilotLinked: true)
  • Show activePrdIds if linked to Task Copilot
  • For agent improvements, parse metadata to extract AgentImprovementMetadata fields
  • Show status counts (pending/approved/rejected) from metadata.status field
  • Truncate rationale to 40 characters for table display
  • For protocol violations, show summary counts by severity
  • Only show violations section if Task Copilot is linked
  • Parse violation context JSON to extract description
  • Format violation dates in YYYY-MM-DD format
  • Highlight critical and high-severity violations

Example Output

## Memory Dashboard

**Initiative:** Framework Improvements v2.0
**Status:** IN PROGRESS
**Task Copilot:** Linked (PRDs: PRD-001, PRD-002)

**Focus:** 4 remaining tasks for v2.0 release
**Next:** Complete CMD-1 memory command

### Recent Decisions
- Migrate to slim initiative format for 75% context reduction
- Use Task Copilot for all task tracking, Memory Copilot for strategic decisions only
- No time estimates policy enforced across all outputs

### Recent Lessons
- Bloated initiatives waste tokens and slow down /continue
- Separating task management from strategic memory improves clarity

### Key Files
- .claude/commands/continue.md
- mcp-servers/copilot-memory/src/index.ts
- docs/operations/working-protocol.md

### Recent Memories (Last 10)
Type       | Content Preview                                    | Created
---------- | -------------------------------------------------- | ----------
decision   | Migrate to slim initiative format for 75% cont... | 2025-01-15
lesson     | Bloated initiatives waste tokens and slow down... | 2025-01-15
decision   | Use Task Copilot for all task tracking, Memory... | 2025-01-14
context    | Framework v2.0 focuses on memory optimization      | 2025-01-14

### Agent Improvements
**Pending:** 2 | **Approved:** 1 | **Rejected:** 0

Agent | Section           | Rationale                           | Created
----- | ----------------- | ----------------------------------- | ----------
me    | Core Behaviors    | Add checkpoint validation before r... | 2025-01-15
qa    | Output format     | Include regression test considerati... | 2025-01-14

**Storage:** ~/.claude/memory/abc123def456/memory.db
**Workspace ID:** abc123def456 (auto-generated)

### Protocol Violations
**Total:** 2 | **Critical:** 1 | **High:** 1 | **Medium:** 0 | **Low:** 0

Type                     | Severity | Description                     | When
------------------------ | -------- | ------------------------------- | ----------
generic_agent_used       | critical | Used "Explore" agent            | 2025-01-15
files_read_exceeded      | high     | Read 5 files (limit: 3)         | 2025-01-15

Additional Information

If the user asks "where is my data stored?", explain:

  • Memory database location: ~/.claude/memory/<workspace-id>/memory.db
  • Archives location: ~/.claude/memory/archives/
  • Workspace ID is auto-generated hash of project path (unless explicitly set)
  • To preserve memories when moving/renaming projects, set WORKSPACE_ID in .mcp.json
  • See Memory Copilot README for details on workspace management

End

Present the dashboard and ask: "Would you like to see more details about any specific memory or initiative field?"