Reviews work artifacts with fresh-context reviewer agents and evidence-based findings.
npx skills add https://github.com/allemaar/open-skills --skill cold-review-allemaar
Please help me install this Agent Skill. Skill: cold-review Source: https://github.com/allemaar/open-skills/tree/main/skills/cold-review Command: npx skills add https://github.com/allemaar/open-skills --skill cold-review-allemaar
High-intent search queries and answers about installing and using this skill.
Invoke cold review on the diff or files and state the objectives. The skill spawns fresh reviewer agents with distinct lenses such as correctness, security, and tests, then synthesizes their evidence-backed findings into one scored report.
One reviewer for narrow work like a single file, two for medium mixed work, and three for broad or high-risk work such as security changes or migrations. More than three reviewers requires explicit user confirmation.
Use cold review for outside critique of concrete work artifacts. Use verify for a formal self-gate on intent, plan, or execution, and double-check when the current agent should re-read and challenge one specific target itself.
The skill performs a degraded single-agent review in the same session and clearly labels the report as degraded, warning that it may miss context-bias issues that fresh reviewers would catch.
No. The skill explicitly states that a high review score does not replace tests, builds, or formal verification. It is a critique layer, not a substitute for executable validation.
materkeyReview feature-level architecture in plan files or implementation diffs for design risks.
vTRKAScore project artifacts against a confidence rubric and return PASS/WARN/BLOCK JSON results.
sizukutamagoRun test suites and generate a formal verification report.
SujinHwang27Design adversarial reviewer personas that critique creation skill outputs for edge cases.
