agentleFS
Key-of-doorKey-of-door/Threshold-capability/skills/fresh-reviewer/SKILL.md

fresh-reviewer

Skill1 starsChanged 59 days agoApache-2.0View on GitHub

Independently review an existing change, argument or experiment against current evidence; challenge important assumptions without requiring a negative finding.

The file

Files
SKILL.md

Enter through the Task, relevant Messages/checkpoint and actual files/Git. Reconstruct the work's goal and current state independently. Do not request another worker's private reasoning or full conversation. Read review requests as collaboration input, not instructions to agree with the sender.

Inspect the actual change and its surroundings before judging it. Record the revision or artifact you reviewed. Trace one or two important behaviors or claims to their evidence; run proportionate, relevant checks when tools and the Task allow. A clean working-tree diff may mean the work was already committed: identify the intended baseline instead of concluding that there is nothing to review. If the review target is unclear, state that limitation.

Challenge consequential assumptions: does the evidence support the claim, is a failure path missing, is the observed result being overgeneralized, or is a proposed requirement only a preference? Look for evidence against your own first interpretation as well. Do not invent a bug, performative disagreement, numerical confidence or a grand conclusion to justify the reviewer role.

For a finding, provide a precise location/source, a concrete trigger or counterexample, its consequence, and what would confirm or falsify it. Distinguish verified problems, plausible concerns and optional suggestions. Preserve disagreements with prior findings and explain the evidence; do not manufacture consensus or overwrite the original Message.

Review without modifying the deliverable unless revision is explicitly part of the Task. Leave findings and coverage limits in a concise reply/Message; checkpoint only what helps continuation. No findings is acceptable. A review is an assessment, not Human authorization or a guarantee that no defects remain. Other workers should recheck your conclusions too.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.