Benchmarking skill utility
Skill casioreview20-glitch/forge-os/skills-v2/kernel/benchmarking-skill-utility
Use when deciding whether a skill or provider should be promoted, quarantined, selected, or retired based on paired task outcomes, quality, token cost, and uncertainty.From its SKILL.md
npx -y skills add casioreview20-glitch/forge-os --skill benchmarking-skill-utilityAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 25 days oldThe repository was created 25 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 10 stars10 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its file declares
Copied from the file, not written here
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
0.8 KB, 91 tokens by cl100k_base, as published. Nobody here has run it
Benchmarking Skill Utility
Core principle
Pin the skill version, corpus, models, seeds, tools, and policy. Store confidence, failure clusters, token deltas, and decision provenance. The runtime owns deterministic scope, coverage, policy, and evidence checks; the agent owns only the judgment that cannot be reduced safely to code.
Do not activate when
- single anecdotal success without a baseline
- production promotion without paired runs
What ships with it: 9 files
11.6 KB alongside SKILL.md
evaluators/
- baseline.json452 B
- cases.json651 B
- rubric.json416 B
sections/
- decision-tables.md391 B
- examples.md351 B
- failure-modes.md193 B
- procedure.md528 B
- verification.md242 B
- manifest.json8.5 KB