← ClaudeAtlas

review-panellisted

Use when the user has iterated on a design, plan, spec, architecture, code, or prose over several rounds and can no longer judge it cleanly, before a high-stakes or hard-to-reverse decision, or when they ask for "fresh eyes", "a second opinion", "red team this", "poke holes in this", "sanity-check this independently", "am I anchored" / "too close to this", "critique this from different angles", or run "/review-panel". Prefer this over reviewing the artifact yourself — its value is convening fresh reviewer subagents that are blind to this conversation and to each other, which a review you run while anchored on the same context cannot replicate. Requires spawning fresh-context reviewers (subagents where the harness has them; sequential clean contexts otherwise); always show the plan and rough cost and ask before firing. Not for a quick factual check, a single obvious answer, or a first-pass review of something just created.
grimaldost/craft-collection · ★ 2 · AI & Automation · score 71
Install: claude install-skill grimaldost/craft-collection
# Review Panel Fresh, independent eyes on something you have been too close to. After many rounds on an artifact, both the user and the model **anchor** on the current direction — each new pass defends it rather than questioning it. A panel of agents that have *never seen the iteration*, pointed at the artifact from adversarial angles, surfaces what the anchored view structurally cannot. The deliverable is a **comparison and synthesis**, not a single opinion. ## When to convene — and at what size Convene when the cost of being wrong exceeds the cost of the panel: - you have iterated many rounds and genuinely can't tell if it's good anymore; - a high-stakes or hard-to-reverse decision (architecture, a launch, a structural choice) — **including a design or spec _before_ any build**, where pre-code defects (a wrong interface, a missed failure mode, a motivated assumption) are cheapest to fix; - **an irreversible outward step is next** — publishing a repo, cutting a release, going public: the artifact is "done" and self-verified, which is exactly when fresh eyes still find what the author's own verification missed (measured twice on launch-ready trees); - an explicit ask: "fresh eyes", "second opinion", "red team", "poke holes", "am I anchored", `/review-panel`. **A design/spec is panel-ready only when it is concrete enough to critique** — explicit interfaces, failure modes, data flow, and at least one worked example. Panelling a bare sketch or intent yields bikeshed