skill-testing

Playwright for coding agents. Test that your skills, MCP servers, and CLIs actually work when an agent uses them — sandboxed YAML suites, A/B experiments, CI gates.

skilled: open-source agent skill evaluation CLI and MCP server for Codex and Claude Code. Test skills, detect regressions, and improve with evidence.

Tests and scores any Agent Skill against the official anthropics/skills specification. Use this skill when you need to check if a skill repository or SKILL.md file is compliant with the Agent Skills standard, audit skill quality, get a compliance score, or receive specific improvement suggestions.

Quality and trigger-evaluation gate for reusable AI agent skills.

Automated end-to-end testing for AI agent skills(agentskills.io). Launches Claude Code and Cursor as subprocesses, runs scenarios in real workspaces, and asserts what the model actually does.

B/O/E evaluation framework for SKILL.md-based LLM operating systems and agent hubs

Validate, pressure-test, and release-gate SKILL.md packages for OpenAI and portable agent runtimes.

Behavioral test harness for Claude Code skills — run a skill in an isolated fixture and assert on what it actually does

CLI runner for testing AI coding-agent Skills with sandboxed workspaces, deterministic assertions, provider adapters, and CI-ready reports.

Claude Code skill that tests if agents follow rules, checks tool order, and writes reports. Built on affaan-m/ECC by @affaan-m.

A Claude Agent Skill validator for Claude Code: spec conformance, script testing, quality and security scoring, a Mermaid skill map, and per-skill token cost.

Test and validate OpenClaw skills locally before publishing to ClawHub — checks triggers, env vars, scripts, and SKILL.md completeness without needing a live agent. Zero dependencies.

Testing harness for AI agent skills and plugins — contract-driven YAML specs with mocked environments

Vet a Claude Skill before you trust it: read it as code, then test its behaviour in a sandbox

你的 Skill 到底有没有用？跑一次就知道——用真实任务做对照测试，看它在每个模型上是帮忙还是拖后腿。 | Measure whether a skill actually helps, per model, via paired with/without experiments and blind judge review.

Automated QA system for Claude Code skills. Discovers skills, generates tests, captures screenshots, scores quality (A-F grades), and publishes an interactive dashboard.

Agent Skill Infrastructure: quality check, behavior test runner, version awareness

Evergreen: self-maintaining skills, docs and codemaps for AI agents (research refresh, learnings, evals, git self-publishing)