code-review plugins

by Han

Refactor code using the Refactoring Guru / Fowler catalog: detect code smells, apply the matching named refactoring in small behavior-preserving steps, verify after each.

by Anna Rosen

Research-coding workflow for computational science, built with a JAX/Python astrophysics group and domain-agnostic in its contracts. The researcher owns the science: scientific assumptions are proposed and approved before they become load-bearing, experiments carry a derived prediction, and validation is staged (physics and analytic checks in development, published-result or reference-code comparison before release). 43 skills: model development and ideation, literature, scoping, building correctly, equation-critical sources, verification and numerical validation, Bayesian inference, code and figure review, performance, records and provenance, MyST documentation, reproducibility and release. Plus an equation-verifier agent, six slash commands, twelve self-limiting hooks (Stop gates warn by default; RWF_STRICTNESS=standard blocks), and per-domain lenses for reference-parity work. The same skill files load in Codex via a symlink.

by lab1702

Run independent whole-repository reviews, fix verified issues, run checks, and make authorized local commits until two consecutive passes are clean.

by Matheus Malagris

Single review-only entry point for an existing diff — runs verify → code-review → implementation-review (and security-audit only when the change touches a trust boundary), then consolidates every finding into one go/no-go verdict.

by yukineko

v0.1.24 integrates this line with the remote's. The review is now scoped to what THIS session edited: `harness_core::transcript::files_edited_by_session` reads the session transcript and `review::attribute` narrows the changed-file list to it, so a peer session's uncommitted work in a shared tree is no longer presented as your diff. Attribution is three-valued — when the transcript cannot be read the list is NOT narrowed and the block says so, rather than silently reviewing nothing. Also: the session-scoped skip now carries a third state. Session-scoping fixed WHOSE token is consumed; it did not fix WHEN. A Stop is adjudicated by four independent gate processes, so a token honoured by one that allows was burned even when another blocked the same stop — the stop never happened and the one-shot escape had been spent on nothing, which is standing pressure toward a permanent bypass (CLAUDE.md 5). `consume_session_skip` now takes `stop_hook_active` and parks an honoured token as `<id>.skip.honoured`: it is re-honoured while a re-entry after a block is in flight, and deleted the first time a chain actually ends. A rename failure that can neither bound nor clear the token is resolved to NOT honouring it. Code-review gate for Claude Code: on Stop, review the diff before the agent can declare done. In inject mode it blocks the stop and injects a review rubric so the running agent self-reviews (no API key); in subprocess mode it runs an independent reviewer and surfaces only its findings. Subscription-native, bundled Rust binary. v0.1.21: the shared project-root `.reviewgate-skip` marker is gone. The operator escape is now `reviewgate skip --reason "..."`, which is scoped to the issuing session (CLAUDE.md 5 forbids a shared one-shot file under parallel sessions, since whichever session stops next consumes it), requires a non-empty reason, and appends both the issue and the consumption to the gate log so a bypass cannot happen unrecorded.

by Skylar

Verification-first workflows for finding real bugs, closing inconsistencies, and making measured improvements. Prove-it-or-don't-report-it: every finding carries an evidence tier and a proof artifact, fixes ship a failing-then-passing regression test fixed at root cause, and inconsistencies are closed with a mechanical enforcement. v2 adds test-suite validation (flaky + mutation testing), characterization safety nets, root-cause sibling sweeps, regression bisection, and a regression guard.

by Skylar

Verification-first workflows for finding real bugs, closing inconsistencies, and making measured improvements. Prove-it-or-don't-report-it: every finding carries an evidence tier and a proof artifact, fixes ship a failing-then-passing regression test fixed at root cause, and inconsistencies are closed with a mechanical enforcement. v2 adds test-suite validation (flaky + mutation testing), characterization safety nets, root-cause sibling sweeps, regression bisection, and a regression guard.

by vickoman

Evidence-driven Research, Plan, Implement workflow for controlled code changes in Codex.

by Rust Studio contributors

A maintainer-grade Rust engineering studio with 62 focused skills for architecture, implementation, review, testing, performance, security, and release work.

by Matheus Malagris

High-confidence security review of a diff, layered on wstg-security-testing. Confidence gate, false-positive precedents, WSTG-ID mapping, and a dependency audit. Reports only concretely exploitable findings.

by OpenAI

Scans authorized code to detect and confirm plausible vulnerability findings. Supports codebase investigation, change set analysis, and minimal fix preparation.

by dimonb

Drive one change end-to-end through a repo's pipeline — issue, spec, implementation, its own review passes, hand-off or merge — on GitHub or GitLab. Self-driven: every review pass runs as read-only subagents with adversarial verification, and nothing waits on an external reviewer.

by Andre Paat

Deterministic diff-time detector for swallowed errors and ghost transactions, with a global PreToolUse commit gate that hands findings to the current Codex session to review.

by SimpleCORE Inc.

Development handbooks for the SimpliX stack — Spring Boot backend conventions with YAML-driven code generation, simplix-react frontend conventions with OpenAPI-driven scaffolding, and a browser-driven end-to-end usability audit

by Dario Amato

Catches bugs before they ship, and blocked features before they're built: per-task model/effort routing, stack-aware TDD (Java, Angular, React/Next.js, Python), and a two-pass adversarial reviewer (Dylan Dog). A fork of obra/superpowers.

by Software Factory contributors

Jstack delivery, architecture, evidence, review, and writing skills for production software work.

by NSRandom

Staged, multi-model build pipeline: plan, specify, implement, review, validate, debug.

by Oleksandr Hiliazov

Six process skills that enforce spec-before-plan, plan-before-code, TDD, root-cause debugging, and evidence-before-claim. Stack-agnostic: per-project commands come from a .claude/subpowers.md contract.

by Jesse Vincent

An agentic skills framework & software development methodology that works: planning, TDD, debugging, and collaboration workflows.

by korya

Software development skills: product specs, deep planning, root-cause analysis, agent-oriented repo docs, spec-aware rebasing, PR review, PR submission, review revision, and end-user e2e testing.

by Luiz Tadeu Mendonça

The knowledge layer of a public agent-harness reference: 15 dense, opinionated engineering skills covering AWS infrastructure, React/Vite frontends, BFF-on-Lambda backends, CI/CD, the development loop and its quality gates. Written project-agnostically with <project> / <apex-domain> placeholders. This Power ships skills only — the harness's personas, permission hooks and merge gates are deliberately not exported.

by Jack Blacketter

The tagteam handoff contract for Claude — the /handoff skill and a SessionStart hook that surfaces the current cycle. The engine is the tagteam CLI (uv tool install tagteam).

by erstaples

Walks through a diff with editor navigation and a persistent, evidence-backed living change dossier.

by Andre Paat

Multi-provider code review — Codex + Grok + Claude (diff-only) by default; Gemini, OpenCode DeepSeek, OpenCode GLM and Qwen opt-in, with calling-context arbitration