🪞

Self-Reflection

essential

Agent self-improvement through introspection

Turn vague dissatisfaction into specific, technical changes. The agent scans its own system — persona, skills, config, cron jobs — diagnoses the root cause, and makes concrete improvements.

Instructions for your agent

# Self-Reflection Skill

When the owner wants to improve how you operate, follow this process. The goal: turn vague dissatisfaction into specific, technical changes to your own system.

## Process

### 1. Understand the Problem (2-3 questions max)
Ask focused questions to pin down:
- **What** specifically is wrong? (Get a concrete example)
- **What would "good" look like?** (Expected vs actual behavior)
- **How important is this?** (Tweak vs fundamental change)

Don't over-ask. If the complaint is clear enough, skip to step 2.

### 2. Deep System Scan
Audit ALL relevant parts of your system. Read before changing:

#### Core Identity & Behavior
- Persona, boundaries, tone rules
- Operating instructions
- User profile, preferences
- Long-term memory and facts
- Tool-specific notes

#### Skills (behavior patterns)
- All custom skills
- Bundled skills
- Workspace skills override bundled on name collision

#### Configuration
- Model, tools, channels, heartbeat, queue mode, streaming, etc.

#### Cron Jobs
- All scheduled jobs (timing/frequency related issues)

**Read broadly, change surgically.** Scan everything related, modify only what's needed.

### 3. Diagnose & Propose
Present findings:
1. **Root cause** — what causes the unwanted behavior
2. **Proposed changes** — specific files and edits
3. **Side effects** — anything else affected
4. **Alternatives** — multiple approaches if they exist

### 4. Implement Changes
After approval, make all changes:
- Edit workspace files (persona, memory, etc.)
- Edit/create/modify skills
- Update config
- Add/modify/remove cron jobs

**Every change must be technically concrete.** "I'll be more careful" is NOT a valid change.

### 5. Verify & Document
- Test the change if possible
- Document what changed and why
- Commit workspace changes

## Important Principles
- **Scan everything, change only what's needed**
- **No fake fixes** — if you can't find a technical change, say so
- **Owner approves changes** — present the plan, get approval, then execute
- **Compound improvements** — each reflection makes the system permanently better
- **Be honest about limitations** — some behaviors are model-level