Run verification commands and write a structured verdict to build-eval.json.
npx skills add https://github.com/danielkhigay/gaz --skill evaluate-danielkhigay
Please help me install this Agent Skill. Skill: evaluate Source: https://github.com/danielkhigay/gaz/tree/main/.claude/skills/evaluate Command: npx skills add https://github.com/danielkhigay/gaz --skill evaluate-danielkhigay
Evaluates a completed phase step by executing verification commands, orchestrating review agents, and producing a structured verdict written to build-eval.json. It does not fix issues — it surfaces failures for follow-up review.
Invoke the skill by specifying the phase and step to evaluate, and it will run verification commands, spawn agents, and write the verdict.
High-intent search queries and answers about installing and using this skill.
To automate phase step verification, specify the phase and step to evaluate, and the system runs verification commands, spawns specialized QA agents, and writes a structured verdict to build-eval.json.
When a builder signals ready_for_eval, quality verification triggers automated command execution and spawns specialized QA agents to review changes and synthesize evidence into a build-eval.json verdict.
Specialized QA agents review changes and synthesize evidence to produce a final verdict, writing the structured evaluation record to build-eval.json for traceability.
Ad-hoc quality verification can be applied across project phases to execute verification commands and spawn agents, surfacing failures for follow-up review without fixing issues.
Automated verdicts surface failures for follow-up review without fixing issues, meaning builders must manually address any identified problems after the structured build-eval.json verdict is produced.
baleen37Enforce minimal Claude commands that delegate complex workflows to skills.
danielkhigayExecute a single Gaz blueprint phase step and update build-eval.json.
thrawn01Coordinate subagents to execute multi-phase plans sequentially with verification.
samuelcastroVerifies AI engineering output claims via adversarial sub-agents producing structured verdicts with confidence scores.
