# Ezra Monitor

> Ezra Monitor

- Skill: `letta-ai/ezra-monitor-2` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add letta-ai/ezra-monitor-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/letta-ai/ezra-monitor-2/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: letta-ai (https://skillmd.com/u/letta-ai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/letta-ai/ezra-monitor-2

---

# Ezra Monitor

Monitor and examine ezra-prime's activities to propose memory block updates.

## Purpose

Self-reflection tool that analyzes ezra-prime's Discord support conversations to:
- Identify recurring patterns and issues
- Detect knowledge gaps from repeated questions
- Propose updates to memory blocks based on learnings
- Track team corrections and user feedback

## Scripts

### `scripts/fetch-recent.sh`
Fetches recent conversations and messages from ezra-prime.

```bash
./scripts/fetch-recent.sh [hours] [limit]
# Default: last 24 hours, 50 messages
```

### `scripts/analyze-patterns.sh`
Analyzes fetched messages for patterns.

```bash
./scripts/analyze-patterns.sh <messages.json>
# Outputs: recurring topics, unanswered questions, corrections
```

### `scripts/propose-updates.sh`
Generates proposed memory block updates based on analysis.

```bash
./scripts/propose-updates.sh <analysis.json>
# Outputs: suggested edits to specific memory blocks
```

### `scripts/scan-corrections.sh`
Scans conversations for user/team corrections that need action.

```bash
./scripts/scan-corrections.sh [agent_id] [num_conversations]
# Default: ezra-prime, 10 conversations
# Outputs: corrections.json with HIGH/MEDIUM/NORMAL priority items
```

Detects:
- Team corrections (cameron, swooders, pacjam, 4shub) - HIGH priority
- User corrections ("that's wrong", "actually", etc.)
- Link shares after responses (user providing sources)

## Environment

Requires:
- `LETTA_API_KEY` - API access
- `EZRA_PRIME_AGENT_ID` - Target agent (default: agent-57ce3ea1-72ad-43e5-a444-7e3724f706e8)

## Workflow

1. Run fetch-recent.sh to download conversation data
2. Run analyze-patterns.sh to identify learnings
3. Run propose-updates.sh to generate memory edit suggestions
4. Review and apply updates manually or via API