Review Code on Two Axes: Standards and Spec in Parallel

Where this fits

A developer asks "can you look at my PR?" They don't want a generic "looks good." They want two answers: Does this follow our coding standards? Does it implement what the issue actually asked for? Manually context-switching between style rules and issue acceptance criteria is slow. The code-review skill splits those concerns into two parallel review agents and reports them side by side.

Why agents benefit

• Parallel sub-agents: Style and spec reviews run in isolated contexts, so style findings don't drown out correctness gaps, and each agent only sees relevant evidence.
• Fixed-point diffing: Handles commits, branches, tags, or merge-base, so "review since X" always has a precise diff without guessing.
• Repo-aware standards: It looks for documented coding standards in the repo, not generic best practices, making suggestions actionable.
• Spec traceability: The spec agent checks changes against the originating issue/spec, flagging missing or extra behavior.
• Combined report: Aggregated side-by-side output gives both style and spec findings in one pass, saving the developer a second review loop.

Practical scenarios

WIP branch review before opening a PR

Developer: "Review since main." The skill pins the diff, standards agent checks naming and formatting, spec agent checks the partial implementation against the issue. Findings surface before CI runs.

PR review when the repo has coding standards

Repo with CONTRIBUTING.md or eslint config? Standards agent verifies compliance. Spec agent confirms behavior matches linked issue.

"Review since tag v2.1" for a hotfix

Developer wants to know what changed between release tag and current branch. Skill uses tag as fixed point, giving a clean review of only the hotfix diff.

Add it to your agent workflow

Install with:

npx skills add ArcReel/ArcReel --skill code-review

Then invoke naturally:

User: Review the changes since origin/main against the issue tracker.
Agent: Pinned diff origin/main..HEAD.
Running standards review and spec review in parallel...

Make sure docs/agents/issue-tracker.md exists; if not, run /setup-matt-pocock-skills.

Compare before adopting

What to check before adding it to your stack:

• Quality signals: How specific are its findings? Does it cite the repo's actual standards and issue requirements?
• Maintenance freshness: Is the skill actively updated? Review the repo's commit and issue activity.
• Alternatives: Compare with single-pass review skills or generic code-review prompts. The two-axis split is the key differentiator.
• Workflow fit: Does it integrate with your agent's existing issue-tracker setup? It expects docs/agents/issue-tracker.md.

Why it is worth tracking

The two-axis architecture is a strong pattern: isolating context per concern often yields better review quality than one long agent pass. At 4,310 stars with an active ArcReel repo, the skill is gaining community validation. If your agents already handle PRs, tracking it is low-effort and high-signal. For a live index of this skill and related options, see https://www.openagentskill.com/skills/arcreel-code-review.