@meir-labs/skill-tool-eval
v0.1.1
Published
The meirlabs tool-picking skill — a multi-agent evaluation harness: verified pricing research, adversarial fact-checking, a judge panel on the buyer's real priorities, and a decision memo with switch triggers. npx installs it into .claude/skills.
Maintainers
Readme
@meir-labs/skill-tool-eval
The meirlabs tool-picking skill — runs "what platform should I use for X" as a multi-agent evaluation instead of a from-memory answer: parallel research with pricing verified against official pages, adversarial fact-checking of the decision-flipping claims, a judge panel scored on the buyer's real priorities, and a decision memo with a runner-up and switch triggers.
Install
# into the current project (./.claude/skills/tool-eval)
npx @meir-labs/skill-tool-eval
# into every project (~/.claude/skills/tool-eval)
npx @meir-labs/skill-tool-eval --globalPass --force to overwrite an existing copy. Claude Code picks the skill up automatically when asked to choose a tool, platform, or vendor where the decision has to hold up long term.
The skill is plain markdown — read it before you run it: files/.
Full page: meirlabs.com/skills/tool-eval.
