Skip to content

runkids/agents-context-router / .skillshare/skills/agents-context-router-code-review

agents-context-router-code-review

Verified

Review an agents-context-router diff, branch, commit or PR for correctness, repository rules, requested behavior, affected consumers and cleanup. Use when reviewing changes or preparing a commit, push or pull request. Read-only unless fixes are requested.

Review one change in five independent passes. Read docs/review.md for this repository's consumers and verification commands; load docs/releasing.md when versions, manifests or workflows change. These docs own the repository rules.

1. Pin the scope

Inspect Git status and preserve unrelated edits. Refresh the remote base when reviewing against it, then record the base SHA, head SHA and uncommitted paths. With no target, review git diff origin/main...HEAD plus git diff HEAD and task-owned untracked files. With a ref, use that ref as the base. For a PR, retrieve its current head and review that code in an isolated checkout, not the working branch. A path argument means review that file or directory whole.

Locate the originating issue, PR description or user request. If none is available, report "no spec" for that pass rather than inventing requirements. Trace definitions, callers, tests and every changed public concept before reporting a finding.

2. Run five passes

The passes are separate so that one pass's conclusions do not steer another. For a diff that touches several consumers, run them as independent subagents when available and give each the pinned diff, spec and relevant repository docs. For a small diff, run the passes yourself in sequence: each subagent re-reads the same context, which costs more than it adds.

  1. Correctness: wrong topic output, missed checks, broken errors, path escapes, symlink handling, encoding and Python 3.8 compatibility. Preserve fail-closed behavior and keep version output independent of router configuration.
  2. Rules: documented scope, version agreement, release authorization, native platform contracts and verification requirements. Cite the actual rule.
  3. Spec: missing or partial requirements, unintended behavior and scope expansion. Tie each finding to the request or spec.
  4. Consumers: sweep the source script, routed fixture, templates, skill, README locales, native manifests, release configuration, workflow and tests for every changed concept. Report consumers that were missed.
  5. Cleanup: duplicate mechanisms, unused code, extra dependencies or options, and fixes placed in one caller instead of shared code. Label judgment calls.

Each proposed finding needs a file and line plus a concrete input or state that produces the wrong result. Style preferences alone are not findings.

3. Verify and report

Read every cited location and confirm the scenario, using the narrowest relevant check where practical. Mark findings confirmed or plausible and state what remains unknown; discard disproved and duplicate findings.

Report findings separately by pass, followed by executed checks and limitations. Distinguish pre-existing issues from regressions. If fixes are authorized, apply only confirmed in-scope fixes, add a focused regression check where practical, and rerun the affected checks. Commit, push, public comments and publication require the user's authorization; a review request alone is read-only.

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 469kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall