Agent skills that work across coding agents.
| Agent | Command |
|---|---|
| Pi | pi install https://github.com/sheurich/agent-skills |
| Claude Code | /plugin marketplace add sheurich/agent-skills |
| Gemini CLI | gemini extensions install sheurich/agent-skills |
| npx skills | npx skills add sheurich/agent-skills --all -g |
Install a single skill with npx:
npx skills add sheurich/agent-skills -s <skill-name> -g| Skill | Description |
|---|---|
| readiness-review | Structured readiness assessment before proceeding to the next environment or phase. |
| semgrep-scan | Scan changed files with Semgrep for security and correctness issues before committing. |
| skill-authoring | Write and refine agent skills. Use when generating skills from docs, extracting from sessions, designing tool interfaces, managing lifecycle, or writing for dual-harness patterns. |
| swival | Delegate tasks to Swival for self-reviewed code changes, secret-safe operations, cached analysis, and A2A agent serving. |
| Plugin | Description |
|---|---|
| auto-summarize | Rolling session summary and automatic session naming for Pi. |
See CONTRIBUTING.md for how to add skills, write tests, and run validation.
skills/ # Simple skills (SKILL.md only)
plugins/ # Full plugins (own manifests, hooks, LSP, MCP)
tests/scenarios/ # One scenario per skill
scripts/validate-extra.sh # Repo-specific validation checks
vendor/agent-validate/ # Vendored validator
validate.sh # Thin wrapper → vendor/agent-validate
.claude-plugin/marketplace.json
gemini-extension.json
./validate.sh # run all checks
./validate.sh --skip json # skip specific checksCI runs agent-validate on every push and pull request.