skillbench
Track skill versions, benchmark performance, compare improvements, and get self-improvement signals. Integrates with tasktime and ClawVault.
- License 许可
- MIT
- 最近更新
- 2026-06-07
- Stars
- ★ 0
- Forks
- 1
01 前言 Front Matter
- name
- skillbench
- metadata
- { "openclaw": { "install": [ { "id": "node", "bins": [ "skillbench" ], "kind": "node", "label": "Install SkillBench CLI (npm)", "package": "@versatly/skillbench" } ], "requires": { "bins": [ "skillbench" ] } } }
- description
- Track skill versions, benchmark performance, compare improvements, and get self-improvement signals. Integrates with tasktime and ClawVault.
03 SKILL.md 正文
skillbench Skill
Self-improving skill ecosystem for AI agents.
Track skill versions, benchmark performance, compare improvements, and get signals on what to fix next.
Part of the ClawVault ecosystem | tasktime | ClawHub
Installation
npm install -g @versatly/skillbench
The Loop
1. Use a skill → skillbench use github@1.0.0
2. Do the task → tt start "Create PR" && ... && tt stop
3. Record result → skillbench record "Create PR" --success
4. Check scores → skillbench score github
5. Improve skill → Update skill, bump version
6. Repeat → Compare v1.0.0 vs v1.1.0
Commands
Track Skills
skillbench use github@1.2.0 # Set active skill version
skillbench skills # List tracked skills + signals
Record Benchmarks
# Auto-pulls duration from tasktime
skillbench record "Create PR" --success
# Manual duration
skillbench record "Create PR" --duration 45s --success
# Record failures
skillbench record "Create PR" --fail --error-type "auth-error"
Score & Compare
skillbench score # All skills with grades
skillbench score github # Single skill
skillbench compare github@1.0.0 github@1.1.0
Export & Dashboard
skillbench export --format markdown
skillbench export --format json
skillbench dashboard # Generate HTML dashboard
skillbench dashboard --open # Generate and open in browser
Automated Testing
skillbench test tasktime@1.1.0 # Run smoke test
skillbench test tasktime@1.1.0 --suite full # Run named suite
skillbench test tasktime@1.1.0 --dry-run # Test without recording
Sync
skillbench sync --clawhub # Import installed skills
skillbench sync --vault # Sync to ClawVault
skillbench sync --all # Everything
Health & Monitoring
skillbench health # Overall health report with alerts
skillbench watch --once # Run all test suites once
skillbench watch --interval 300 # Continuous monitoring every 5 min
Analysis & Improvement
skillbench improve # Get suggestions for weakest skill
skillbench improve github # Improvement plan for specific skill
skillbench trend tasktime --days 30 # Performance trend over time
skillbench leaderboard # Compare agents (multi-agent setups)
skillbench schedule --interval 60 # Generate cron config for auto-testing
Baselines & Regression Detection
skillbench baseline tasktime --set # Set baseline from current performance
skillbench baseline --list # List all baselines
skillbench baseline --check # Check all baselines (CI-friendly, exits 1 if failing)
skillbench baseline tasktime --remove # Remove a baseline
CI/CD Integration
skillbench ci # Run all tests + baseline checks
skillbench ci --json # JSON output for automation
skillbench badge # Generate shields.io badges for README
Copy examples/github-action.yml for ready-to-use GitHub Actions workflow.
Grading System
| Grade | Score | Meaning |
|---|---|---|
| 🏆 A+ | 95-100 | Elite performance |
| ✅ A | 85-94 | Excellent |
| 👍 B | 70-84 | Good |
| ⚠️ C | 50-69 | Needs work |
| ❌ D | <50 | Broken |
Based on: Success Rate (40%), Avg Duration (30%), Consistency (20%), Trend (10%)
tasktime Integration
When you omit --duration, skillbench pulls from tasktime:
tt start "Create PR" -c git
# ... do work ...
tt stop
skillbench record --success # Duration auto-pulled
ClawVault Integration
Benchmarks sync to ClawVault automatically.
Improvement Signals
skillbench skills shows:
- ⚠️ needs work — Success rate below 70%
- 🕐 stale — No benchmarks in 7+ days
- ↘️ declining — Getting worse over time
Related
04 同仓库更多 Skills
crowd-prompting
A marketplace where AI agents improve prompts, system instructions, tool descriptions, and other text-based content with domain expertise from real-world operations — and earn tokens for valuable contributions.
ceaser
Interact with the Ceaser privacy protocol on Base L2 using the ceaser-mcp MCP tools. This skill uses the ceaser-mcp npm package for ALL operations -- shield, unshield, note management, and protocol queries. All ceaser tool calls use CLI subcommands (npx -y ceaser-mcp <subcommand>). Ceaser is a privacy-preserving ETH wrapper using Noir/UltraHonk zero-knowledge proofs on Base L2 (chain ID 8453).
ceaser-send
Fully automated private ETH transfer via Ceaser Protocol on Base L2 using the ceaser-mcp MCP tools. This skill uses the ceaser-mcp npm package for ALL operations -- all ceaser tool calls use CLI subcommands (npx -y ceaser-mcp <subcommand>). WARNING -- creates observable on-chain links between user wallet and hot wallet (see Privacy Warning). Generates an ephemeral hot wallet with BIP-39 mnemonic recovery, user funds it, agent signs and broadcasts the Shield TX automatically. Exactly one manual step required (funding the hot wallet with ETH). Uses Noir/UltraHonk zero-knowledge proofs.
dilemm-ai
Compete in DilemmAI, the prisoner's dilemma AI arena at dilemm.ai. Use when an OpenClaw agent wants to create an account, design and submit strategy prompts for their bot, enter matchmaking, analyze historical matches and leaderboards, and iterate on bot strategy. Requires headless browser automation (no API available yet).
moldium
Post and manage content on the Moldium blog platform. Triggered by "post to Moldium", "write a blog post", "publish an article", etc.
arxiv-paper-reviews
Interact with arXiv Crawler API to fetch papers, read reviews, submit comments, search papers, and import papers. Use when working with arXiv papers, fetching paper lists by date/category/interest, viewing paper details with comments, submitting paper reviews, searching papers by title, or importing papers from arXiv URLs via API at http://weakaccept.top:8000/.