agentsclimarketplace

Code review

Skill 1415923/ForgeKit/skills/code-review

Review an existing implementation, diff, tests, or prior review findings for correctness, regressions, boundary conditions, and evidence gaps. Use for initial code review, independent checker work, or targeted findings recheck. Do not use to implement a feature, perform a full security audit, or decide release readiness.From its SKILL.md

Install
npx -y skills add 1415923/ForgeKit --skill code-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.0 KB, as published. Nobody here has run it

Code Review

Trigger Boundary

Use this Skill when the user asks to review code or an existing diff, validate implementation evidence, perform an independent checker pass, or recheck named findings.

Do not use it as the maker for a requested implementation. Route a primary security audit to security-review, release readiness to release-check, project takeover to handover-review, and ForgeKit adoption fit to project-suitability. Select only the Skill or Skills justified by the actual intent; these review Skills are not a mandatory pipeline.

Default Mode: Read-Only

Review is read-only by default. Inspect relevant files, diffs, tests, commands, and current change artifacts without modifying source, tests, governance, or status documents. Discovering a defect does not authorize repair.

If the user or governing workflow separately and explicitly authorizes a named repair, end the reviewer role and hand the named finding to a bounded maker workflow. That bounded maker may change only non-empty explicit writable paths, must record changed-path evidence and validation evidence, and must not expand to another finding, path, project, or external action. A later independent recheck returns to read-only mode.

Evidence and Scope

Start with the requested diff or finding set. Load project rules and active change artifacts only when they affect the reviewed contract. Prefer current code and reproducible command or test evidence over plans or narrative claims.

Check correctness, regressions, error and boundary paths, compatibility, data or migration effects, ownership boundaries, and missing tests. Invoke a deeper security review only when the change has an actual security surface. Invoke a release check only for release readiness.

When evidence is missing or cannot be reproduced, report NEEDS_TEST or state that the finding cannot be closed. Do not infer a pass from documentation that merely claims success.

Review Modes

  • Initial review: assess the frozen scope and acceptance contract. Report newly discovered blocking issues only when they expose a real authorization, data, execution, artifact, safety, or false-success failure.
  • Independent checker: remain separate from the maker and verify evidence from a fresh read-only context. Self-review does not satisfy an independent gate.
  • Findings recheck: focus on the named prior findings and classify each as Closed, Partially closed, or Still open. A regression introduced by the fix may still block when it violates the frozen contract or causes a critical consequence. Keep unrelated hardening as follow-up.

A review decision authorizes only the stage named by the governing proposal.

Risk and Checker Boundary

Require an independent checker because of objective impact, explicit user request, or an existing stage, security, migration, or release gate. Do not require one merely because a change touches a particular number of files, lines, or modules.

Output

Lead with findings and cite the closest available file, line, command, or test evidence.

  • BLOCKER: unsafe to proceed or a frozen gate is violated.
  • MAJOR: material correctness, authorization, compatibility, evidence, or regression defect.
  • NOTE: non-blocking hardening, clarity, or follow-up.

Then state open questions, test gaps, the stage-limited decision, and residual risk. If there are no findings, say so and identify any remaining validation gap.

External Actions

Do not commit, push, tag, release, publish, deploy, rotate credentials, change production permissions, or perform another external or irreversible action. Each such action requires explicit user authorization and is never implied by review or fix authorization.

Keep looking

Skills are one crate of 608,475. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.