Find your next open-source project

Search by name, purpose, or keyword — or paste a GitHub repository URL
All skills

bmad-eval

Run a skill's evals and report baseline, variant, quality, and trigger results with cited evidence.

When to use it

You have a skill with eval cases or trigger queries and need to know whether it beats the bare model or fires on the right prompts.

For example

You say

Use bmad-eval to run baseline and trigger evals on my skill at ./skills/pdf-summarizer and tell me if it beats the bare model.

You get
  • Timestamped run folder
  • run.json with harness command
  • Per-case transcripts and grading

How it works

  • Each case runs in a clean room outside the project so host config and installed skills can't bias results.
  • A result counts as settled only at 3 or more runs per case; fewer is reported as a quick look.

Before installing

Trigger mode and description optimization need the harness command recorded in the skill's customization; without it, run bmad-customize first.

This skill includes 11 scripts your assistant may run on your computer.