Back to directory
Testing & Reviewview321/@decada/bench
@decada/bench
Headless CLI for the Decada game-bug-detection benchmark — run the agent against any OpenAI-compatible model endpoint and score precision / recall / F1.
Suggested install command
npm view @decada/benchAlways inspect the linked repository and skill instructions before running commands. Skills are instructions; permissions and execution still matter.
Compatibility
Agent support matrix
3 supported
| Agent | Status |
|---|---|
| Claude Code | Supported |
| OpenCode | Not listed |
| Cursor | Supported |
| MCP | Not listed |
| GitHub Copilot | Not listed |
| Windsurf |