Runtime Self Improvement

persona: name: "Domain Expert" title: "Master of Runtime Self Improvement" expertise: ['Specialized Knowledge', 'Best Practices', 'Industry Standards'] philosophy: "Excellence through expertise." credentials: ['Industry leader', 'Practiced expert', 'Thought leader'] principles: ['Quality first', 'Continuous improvement', 'Evidence-based decisions', 'Customer focus']

Runtime Self-Improvement Skill

Anti-Rationalization Table

When to Use

Trigger phrases:

• "runtime self improvement"
• "Help me with runtime self improvement"

Use cases:

• When the task matches this skill's domain expertise

When NOT to use:

• For tasks outside this skill's scope

Overview

Enable OpenClaw to continuously improve itself at runtime. Monitor performance, detect skill gaps, enhance existing skills, and optimize workflows automatically during operation.

Purpose: Autonomous self-improvement for OpenClaw
Target: 1ai-skills, workflows, prompts, and configurations
Frequency: Continuous during operation

Core Functions

• Primary operation execution with input validation
• Error detection and automatic recovery
• Output formatting and quality assurance
• Integration hooks for downstream consumers

1. Performance Monitoring

During Operation:
- Track skill usage frequency
- Measure success/failure rates
- Monitor response quality
- Log user feedback

2. Gap Detection

Automatic:
- Identify unused skills
- Find skill overlaps
- Detect missing capabilities
- Analyze failure patterns

3. Skill Enhancement

On-Demand:
- Update skill descriptions
- Add new keywords
- Refine prompts
- Improve documentation

4. Workflow Optimization

Continuous:
- Optimize orchestration flows
- Reduce redundant steps
- Add missing integrations
- Streamline processes

Implementation

1. Initialize the skill context with required configuration
2. Load any dependencies or connected services
3. Execute the primary operation
4. Handle errors gracefully with fallback strategies
5. Return structured results for consumption

Hook: After Each Task

// After completing any task, run self-improvement check
async function afterTaskCompletion(task) {
  // 1. Log task metrics
  await logTaskMetrics(task);
  
  // 2. Check for improvements
  const improvements = await analyzeTask(task);
  
  // 3. Apply if significant
  if (improvements.confidence > 0.8) {
    await applyImprovement(improvements);
  }
}

Gap Detection Algorithm

async function detectSkillGaps() {
  // 1. Get all user requests
  const requests = await getRecentRequests();
  
  // 2. Match to skills
  const matched = requests.map(r => findSkill(r));
  
  // 3. Find gaps (unmatched requests)
  const gaps = requests.filter(r => !matched(r));
  
  // 4. Propose new skills
  if (gaps.length > 10) {
    await suggestNewSkill(gaps);
  }
}

Skill Enhancement

async function enhanceSkill(skillName, feedback) {
  // 1. Analyze feedback
  const analysis = await analyzeFeedback(feedback);
  
  // 2. Update skill
  const updates = {
    keywords: [...skill.keywords, ...analysis.newKeywords],
    description: analysis.improvedDescription,
    examples: [...skill.examples, ...analysis.newExamples]
  };
  
  // 3. Apply changes
  await updateSkill(skillName, updates);
  
  // 4. Commit changes
  await autoGitCommit(`improvement(${skillName}): ${analysis.summary}`);
}

Self-Modification Types

This section covers self-modification types for the runtime-self-improvement skill. Key operations include input validation, core processing, and output verification. Refer to the skill overview for detailed usage instructions.

1. Keyword Expansion

Trigger: Skill used but not matched
Action: Add user keywords to skill
Example: User says "fix bug" → add "bug" to debugging skill

2. Description Refinement

Trigger: Skill fails to match
Action: Improve skill description
Example: Add clearer trigger phrases

3. Example Injection

Trigger: Successful task completion
Action: Add to skill examples
Example: Add successful prompt to skill examples

4. Prompt Optimization

Trigger: Repeated failures
Action: Improve skill prompts
Example: Add more specific instructions

Safety Guards

This section covers safety guards for the runtime-self-improvement skill. Key operations include input validation, core processing, and output verification. Refer to the skill overview for detailed usage instructions.

Always Validate

const safeguards = {
  // Don't modify core identity
  protectedFiles: ['SOUL.md', 'USER.md', 'AGENTS.md'],
  
  // Require human approval for major changes
  requireApproval: ['new skill', 'delete skill', 'workflow changes'],
  
  // Limit changes per session
  maxChangesPerSession: 5,
  
  // Always create backup
  backupBeforeChange: true
};

Approval Workflow

async function applyChange(change) {
  if (change.requiresApproval) {
    // Ask human for approval
    const approved = await askHuman(change);
    if (!approved) return;
  }
  
  // Apply with backup
  await createBackup();
  await apply(change);
}

Integration

• Connects with existing toolchain via standard interfaces
• Supports webhook-based event notifications
• Compatible with CI/CD pipelines for automated workflows
• Provides structured output for downstream consumption

With Heartbeat

During heartbeat:
1. Check for skill gaps
2. Analyze recent performance
3. Apply small improvements
4. Log changes for review

With Memory System

Save learnings to:
- MEMORY.md (long-term)
- memory/YYYY-MM-DD.md (daily)
- skill-specific logs

Metrics to Track

Best Practices

This section covers best practices for the runtime-self-improvement skill. Key operations include input validation, core processing, and output verification. Refer to the skill overview for detailed usage instructions.

Do's

✅ Back up before changes
✅ Validate improvements
✅ Log all modifications
✅ Review changes regularly
✅ Test before deploying

Don'ts

❌ Don't modify identity files
❌ Don't delete without backup
❌ Don't change core behaviors
❌ Don't ignore user feedback

Version History

• v1.0 (2026-02-27) - Initial creation

When NOT to Use

• When the task requires domain expertise the agent has not been configured with
• When human review is mandated by compliance or regulatory requirements
• When the task is too trivial to warrant this skill
• When a more appropriate skill exists

Common Rationalizations

Red Flags

• Agent output is not validated against expected quality standards
• Prerequisites are not verified before task execution
• Watch for shortcuts and skipped steps

Verification

After completing this skill, confirm:

• Output meets the defined quality and completeness requirements
• All prerequisites are verified and documented
• All required outputs generated
• Success criteria met

Related Skills

• self-improving - Basic self-improvement
• auto-git-commiter - Auto-commit changes
• skill-performance-monitor - Monitor skill effectiveness

Process

1. Analyze the task requirements
2. Apply domain expertise
3. Verify output quality