Phoenix Loop

Auto-diagnose agent failures, extract reusable recovery patterns, and create local skills to fix recurring blockers while keeping all data private and local.

Rise from failures. Complete tasks persistently.

When the agent encounters blockers, failures, or repeated friction, this skill starts a self-healing loop: Diagnose -> Extract -> Crystallize -> Verify.

Core Mechanism

1. Diagnose

powershell

# Read recent blocked items
Get-Content memory/blocked-items.md | Select-String "Blocker" -Context 3

# Extract failure patterns
Get-Content memory/tasks.md | Select-String "Status: failed" -Context 5

Diagnosis Checklist:

• Is the failure cause clearly stated?
• Were at least 2 solution paths attempted?
• Is minimum unblock input defined?

2. Extract

Extract reusable patterns from failures:

markdown

## Failure Pattern: {pattern_name}
- Trigger: {when_this_happens}
- Root Cause: {root_cause}
- Solution: {fix_steps}
- Verification: {verification_criteria}

3. Crystallize

Write the lesson to a local skill:

text

skills/local/{pattern_name}-recovery.md

Skill Template:

markdown

# {Pattern Name} Recovery

## Trigger
When {condition} happens

## Steps
1. {step_1}
2. {step_2}
3. {step_3}

## Verification
- [ ] {check_1}
- [ ] {check_2}

## Fallback
If failed, execute {fallback_action}

4. Verify

Next time a similar issue occurs:

1. Search skills/local/ for matching skills
2. Execute recovery steps
3. Log result to memory/{date}.md
4. Update skill if needed

Privacy Security

All data stored locally:

• NO external logging of failure data
• NO API keys or tokens in skill files
• NO upload of user task content
• Only pattern names and solution steps recorded
• Skills stored in skills/local/ local directory

Sensitive Data Filter: Before writing to any memory or skill, check and remove:

• apiKey, token, secret, password
• Bearer , sk-, OPENCLAW_
• Personal emails, phones, addresses

Executable Completion Criteria

A phoenix-loop task is complete if and only if:

Criteria	Verification Command
Failure pattern named	Select-String "Failure Pattern" memory/blocked-items.md
Local skill created	Test-Path skills/local/{name}-recovery.md
Skill has trigger section	Select-String "## Trigger" skills/local/{name}.md
Skill has verification steps	Select-String "## Verification" skills/local/{name}.md
Memory updated	Select-String "phoenix-loop" memory/{today}.md
Privacy check passed	Skill file contains no `apiKey

Usage Example

Scenario: Missing API Key Blocks Task

1. Diagnose:

text

Blocker: Missing Brave API key
Attempted: web_search (failed)
Unblock Input: User runs `openclaw configure --section web`

2. Extract Pattern:

text

Failure Pattern: missing-api-key
Trigger: Tool requires unconfigured API key
Solution: 1. Detect missing key 2. Return config command 3. Provide fallback

3. Crystallize Skill:

markdown

# Missing API Key Recovery

## Trigger
When tool returns "missing_*_api_key" error

## Steps
1. Extract required key name
2. Return config command: openclaw configure --section {section}
3. Provide manual fallback option

## Verification
- [ ] User receives clear config instructions
- [ ] At least 1 alternative provided

4. Verify: Next time an API key issue occurs, auto-apply this skill.

Heartbeat Integration

Add to HEARTBEAT.md:

markdown

## Self-Check (Every 24 Hours)
1. Check `memory/blocked-items.md` for blockers older than 24h
2. Run phoenix-loop diagnosis on each long-term blocker
3. If reusable pattern found, create or update skill

Rollback

If the skill causes issues:

powershell

# Disable skill (rename)
Rename-Item skills/local/{name}-recovery.md skills/local/{name}-recovery.md.disabled

# Delete skill
Remove-Item skills/local/{name}-recovery.md

References

• memory/blocked-items.md - Blocker records
• memory/tasks.md - Task history
• skills/local/ - Local skill library

---

Rise stronger from every failure.