EntityProcess The full skill content is bundled with the AgentV CLI and always version-matched to it. Load the specific skill you need. If the task is not about AgentV evals, trace analysis, optimization, or governance, no AgentV skill applies.
agentv skills get <skill-name>
Available Skills
| Skill | Command | Use when |
|---|---|---|
| agentv-bench | agentv skills get agentv-bench |
Run evals, benchmark agents, optimize against evals, compare targets, autoresearch |
| agentv-eval-writer | agentv skills get agentv-eval-writer |
Write, edit, or validate eval YAML files |
| agentv-eval-migrations | agentv skills get agentv-eval-migrations |
Migrate eval YAML across breaking schema changes |
| agentv-eval-review | agentv skills get agentv-eval-review |
Review, lint, or check eval quality before committing |
| agentv-governance | agentv skills get agentv-governance |
Author or lint governance blocks (OWASP, MITRE, EU AI Act, ISO 42001) |
| agentv-trace-analyst | agentv skills get agentv-trace-analyst |
Analyze eval traces, find regressions, inspect tool trajectories |
Quick Start
- Ensure
agentvCLI is on PATH (runagentv --helpto verify) - Pick the skill from the table above
- Run
agentv skills get <skill-name>to load it - Follow the instructions in the loaded skill
CLI Location
If agentv is not on PATH, check:
node_modules/.bin/agentv(project-local install)~/.local/bin/agentv(global user install)- Run from source:
bun apps/cli/src/cli.ts <command>