Run artifact and trigger evaluations for skills in an isolated environment.
npx skills add https://github.com/tiween/tiween --skill bmad-eval-runner-tiween
Please help me install this Agent Skill. Skill: bmad-eval-runner Source: https://github.com/tiween/tiween/tree/main/.agents/skills/bmad-eval-runner Command: npx skills add https://github.com/tiween/tiween --skill bmad-eval-runner-tiween
This Skill provides a clean, isolated environment to run a skill's evaluations, ensuring the results reflect the skill itself and not external factors.
Run the bmad-eval-runner skill with the path to the skill directory: bmad-eval-runner /path/to/skill
High-intent search queries and answers about installing and using this skill.
To evaluate a skill in an isolated environment, use a runner that executes artifact and trigger evaluations separately, ensuring results reflect the skill itself without contamination from global configurations or previous runs.
Skill trigger evaluation measures whether a skill's description triggers Claude on a given query. It runs in an isolated environment to test if the skill activates correctly for the intended inputs.
Yes, you need Python and Claude Code installed to run skill benchmarking. The evaluation process executes scripts that require these dependencies to assess skill behavior and output grading.
To grade skill outputs against expectations, run an artifact evaluation. This process executes the skill against a specific prompt and compares the generated output to predefined expected results.
Inconsistent skill evaluation results often stem from environmental contamination. Running evaluations in an isolated environment prevents global configurations and previous execution states from affecting the skill's behavior.
The best way to benchmark a skill without external interference is to run it in an isolated environment using a dedicated evaluation runner, which separates the execution context from global configurations.
pascallammersCreates isolated Git worktrees with .gitignore safety and automated setup.
pipixia-labsGuide creation of SKILL.md files and coordinate evals, benchmarks, and iteration.
ZeroGravitySkin-RonAutomate Claude Code skill creation with SKILL.md authoring and skill-rules.json configuration.
apassuelloEvaluate LLM outputs with automated metrics, human judgments, and benchmarking.
