# Meta Find Skills

> Discovers community skills when local ones are insufficient, scoring them by credibility and running safety checks before installation.

- Skill: `oyi77/meta-find-skills` (Agent Skill)
- Install (CLI): `npx skillmds add oyi77/meta-find-skills`
- Raw SKILL.md: https://api.skillmd.com/api/skills/oyi77/meta-find-skills/raw
- Safety review: CAUTION (external: skill-scanner PASS, skillspector CAUTION)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Tags: Community Skills, Credibility Scoring, Github Search, Npm, Safety Validation, Skill Discovery
- License: Apache-2.0
- Author: oyi77 (https://skillmd.com/u/oyi77)
- Updated: 2026-08-22
- Page: https://skillmd.com/skills/oyi77/meta-find-skills

---


persona:
  name: "Ada Lovelace"
  title: "The First Programmer - Master of Algorithmic Discovery"
  expertise: ['Search', 'Pattern Matching', 'Credibility Analysis', 'Safety Validation']
  philosophy: "The analytical engine can do whatever we know how to order it to perform. My mission - ensure no capability gap goes unfilled."
  credentials: ['First computer programmer', 'Mathematical visionary', 'Pioneer of algorithmic thinking']
  principles: ['Search locally first', 'Validate before trusting', 'Score by merit not popularity', 'Safety is non-negotiable']

# Find Skills - Intelligent Skill Discovery System

## Overview

Automatically discover and integrate community skills when your local skills dont cover a need. Works as the **discovery layer** of the self-evolving system - before creating new skills always check if they already exist.

**Makes your AI agent complete and self-sufficient** - never say "I cant do that" again!


## Anti-Rationalization Table

| Rationalization | Reality |
|---|---|
| "I'll figure it out as I go" | A structured approach saves time and reduces errors. Follow the workflow in this skill rather than improvising. |
| "I already know this topic" | Familiarity breeds shortcuts. Use the checklist to verify you haven't missed critical steps. |
| "This doesn't apply to my situation" | The patterns here generalize across contexts. Adapt, don't skip — the underlying principles hold. |
| "One more tool will fix it" | Adding complexity rarely solves process gaps. Master the core workflow first. |

## When to Use
**Trigger phrases:**
- "meta find skills"
- "Automatically discover evaluate and activate community skills when local skills "


**Automatic Activation** when:
- User asks "how do I do X" where X isnt covered by local skills
- User says "find a skill for X" or "is there a skill for X"
- User asks "can you do X" where X is specialized
- The auto-evolve system detects a capability gap
- meta/create-skills checks before generating a new skill

## When NOT to Use

- Local skills already cover the need perfectly
- The request is trivial and doesnt need a dedicated skill
- Youre in an air-gapped environment with no internet

## Skill Discovery Process
1. Validate input and check prerequisites
2. Initialize required connections and contexts
3. Execute core operation with monitoring
4. Validate output against expected format
5. Deliver results and log execution summary


### Step 1: Check Local Skills First

Before searching externally always check whats already installed:

1. Scan all skill activation rules in .opencode/skills/
2. Match user intent against skill descriptions
3. If match found - Use existing skill skip discovery
4. If no match - Proceed to Step 2

### Step 2: Extract Search Intent

Parse the users request into actionable search terms:

- Extract domain keywords (marketing trading devops design)
- Extract action words (create analyze automate optimize)
- Extract platform/context (twitter kubernetes react shopify)
- Form 2-3 search queries combining these terms

### Step 3: Search Community Sources

Query multiple skill registries in parallel:

1. **skills.sh API** - https://api.skills.sh/v1/skills?q={query}
2. **GitHub awesome-openclaw-skills** - Community curated list
3. **npm registry** - Published OpenClaw skill packages
4. **GitHub search** - Public repositories with openclaw-skill topic

### Step 4: Score and Rank Results

Apply credibility scoring algorithm (0-100 scale):

| Factor | Weight | Max Points |
|--------|--------|------------|
| Downloads/Installs | 20% | 20 |
| User Ratings | 20% | 20 |
| Recency (updated within 30 days) | 15% | 15 |
| Author Verification | 15% | 15 |
| Code Quality (lint pass) | 15% | 15 |
| Documentation Quality | 10% | 10 |
| Community Endorsements | 5% | 5 |

**Minimum score to recommend: 70/100**

### Step 5: Safety Validation

Before any skill installation run these checks:

1. **Malware Scan** - Check for obfuscated code eval() exec() patterns
2. **Secret Detection** - Scan for hardcoded API keys tokens passwords
3. **Dependency Analysis** - Verify all dependencies are legitimate packages
4. **Permission Check** - Ensure skill doesnt request excessive permissions
5. **Sandboxed Test** - Run skill in isolated environment before activation

### Step 6: Install and Activate

If skill passes all checks:

1. Download skill files to appropriate category directory
2. Update .skill-activation.json with new rules
3. Verify skill loads correctly in test mode
4. Activate for production use
5. Log installation for meta/performance-monitor tracking

## Integration with Meta-Skills
- 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 meta/create-skills

find-skills searches existing first
- Found? Install existing skill
- Not found? Delegate to create-skills to generate new one

### With meta/auto-evolve

auto-evolve detects capability gap
- find-skills searches for existing solutions
  - Found? Install and activate
  - Not found? create-skills generates new one

### With meta/performance-monitor

performance-monitor tracks discovery metrics
- Query response time
- Installation success rate
- Skill utilization after install
- User satisfaction with discovered skills

### With meta/auto-learner

auto-learner records discovery patterns
- Which queries lead to installs
- Which sources are most reliable
- Which skill types are most needed

## Examples
```
# Basic usage
invoke <skill-name> with appropriate parameters

# Advanced usage with options
invoke <skill-name> --option value --verbose
```


### Example 1: Marketing Skill Discovery

User: "I need to automate my Instagram posting"

find-skills process:
1. Local check: No instagram skill found
2. Search intent: "instagram" "social media" "automate posting"
3. Community search: Found 3 skills
   - social-media-upload (score: 87/100)
   - instagram-automation (score: 72/100)
   - auto-poster (score: 45/100) below threshold
4. Safety check: social-media-upload pass instagram-automation pass
5. Recommend: social-media-upload (highest score)
6. Install and activate

### Example 2: Trading Skill Discovery

User: "Can you help with crypto trading signals?"

find-skills process:
1. Local check: crypto-trading-bot exists but doesnt cover signals
2. Search intent: "crypto" "trading signals" "technical analysis"
3. Community search: Found 2 skills
   - trading-signal-analyzer (score: 82/100)
   - crypto-signals-free (score: 35/100) below threshold suspicious
4. Safety check: trading-signal-analyzer passes
5. Install: trading-signal-analyzer

### Example 3: Gap Detection via Auto-Evolve

auto-evolve: Performance data shows 12 failed requests for "podcast creation"

find-skills process:
1. Local check: No podcast skill
2. Search intent: "podcast" "audio creation" "ai podcast"
3. Community search: No results with score above 70
4. Delegate to create-skills to Generate ai-podcast skill
5. New skill installed and activated

## Configuration

Default configuration can be overridden in config.json:

```json
{
  "apiEndpoints": [
    "https://api.skills.sh/v1/skills",
    "https://raw.githubusercontent.com/openclaw-community/awesome-openclaw-skills/main/index.json"
  ],
  "minCredibilityScore": 70,
  "maxCacheAgeHours": 24,
  "autoActivate": true,
  "safetyChecks": {
    "malwareScan": true,
    "secretDetection": true,
    "dependencyAnalysis": true,
    "sandboxedTest": true
  }
}
```

## Troubleshooting
| Symptom | Cause | Fix |
|---------|-------|-----|
| Operation times out | Network or service issue | Check connectivity and retry |
| Permission denied | Missing credentials | Verify API keys and access tokens |
| Invalid output | Input format mismatch | Validate input against expected schema |


### No skills found
- Broaden search terms
- Try synonyms and related domains
- If still nothing delegate to create-skills

### Skills fail safety check
- Do NOT install - safety is non-negotiable
- Report the skill to community registry
- Generate alternative via create-skills

### Installation fails
- Check network connectivity
- Verify directory permissions
- Try manual installation as fallback

### Found skill doesnt work as expected
- Check version compatibility
- Review skill documentation for requirements
- Report issue to skill author
- Consider create-skills to generate a better alternative

## Common Rationalizations

| Rationalization | Reality |
|---|---|
| "I'll do this later" | Explain why this excuse is wrong for this skill |
| "This is simple, skip steps" | Even simple tasks benefit from process |

## Red Flags

- Meta-skill changes are applied without measuring performance impact
- Agent does not verify that changes maintain backward compatibility
- Watch for shortcuts and skipped steps

## Verification

After completing a skill discovery task, confirm:

- [ ] Search covered 3+ community sources
- [ ] Credibility score calculated for top 5 results
- [ ] Safety validation passed (no known vulnerabilities)
- [ ] Local skills checked first (no duplicate installation)
- [ ] Minimum score 70/100 to recommend installation
- [ ] User confirmed before installing any skill

## Process

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