agentsclimarketplace

Reviewing others code

Skill stevenfackley/opencode-amplifier/.opencode/skills/reviewing-others-code

Contract-governed OpenCode config that amplifies constrained LLMs (Sonnet 4.5, GPT-5.1, cheap corp models) into near-frontier coding agents: multi-agent pipeline with per-agent models, independent test-gen + locked tests, golden-pattern corpus, cross-model review, and an eval harness.

Install
npx -y skills add stevenfackley/opencode-amplifier --skill reviewing-others-code

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

One thing to look at

  • 0 stars0 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.

What its author says it does

Copied from the file, not written here

Use when reviewing someone else's code or pull request (you are the reviewer). Staff-level review discipline — prioritize by severity, separate blocking from non-blocking, review design not just lines, and give kind, specific, actionable feedback that also mentors. Distinct from auto-reviewing your own generated code.

SKILL.md

2.6 KB, as published. Nobody here has run it

Reviewing Others' Code

You are reviewing a teammate's work, not rewriting it. Goal: catch what matters, help it ship safely, and help the author grow — without nitpicking them to death or rubber-stamping.

Read for context first

  • What problem does this PR solve? Read the description + linked issue. Review against the author's INTENT, not your preferred implementation.
  • Skim the whole diff before commenting on any line — understand the shape before the details.
  • Pull and run it if the change is non-trivial. Observed behavior beats read behavior.

Prioritize — comment in this order, and LABEL severity

  1. Correctness — bugs, race conditions, edge/empty/error cases, off-by-ones, wrong logic.
  2. Security — injection, authz/authn, secrets in code, unsafe deserialization, input validation.
  3. Design / architecture — wrong abstraction, leaky boundaries, decisions that are hard to reverse, missing tests for the risky part. This is where staff review adds the most value.
  4. Maintainability — clarity, naming, duplication, complexity that will bite in 6 months.
  5. Style / nits — last, and only if a linter doesn't already own it.

Label every comment [blocking], [consider], or [nit]. The author must never have to guess what must change versus what's optional.

How to write each comment

  • Specific: file:line, the concrete problem, and a suggested direction — not just "this is wrong."
  • Ask when you might lack context: "what happens if x is null here?" beats a wrong decree.
  • Explain the why — that's the mentoring; a reason teaches, a demand doesn't.
  • Call out genuinely good choices. Review isn't only fault-finding.

Calibrate to author and change

  • A hotfix and a greenfield feature deserve different bars — don't gold-plate an urgent fix.
  • Junior author → more teaching + links; senior author → trust more, focus on the hard parts.
  • Team conventions and the linter win over your personal style. Don't dress up taste as correctness.

Close with a clear verdict

Approve / Approve with nits / Request changes, plus a 2–3 line summary: what's strong, what's blocking, what's optional. Respect the author's time and their ownership of the code.

Gives 0 of the 12 instructions most review quality skills give

Counted across 1,048 of the 1,783 authors here whose files we hold, read 2026-08-06

  • ask questions one at a timein 82 of 1048, across 54 files
  • provide a recommended answer for each questionin 73 of 1048, across 45 files
  • explore the codebase instead of asking answerable questionsin 66 of 1048, across 37 files
  • resolve dependencies between decisions one-by-onein 42 of 1048, across 15 files
  • interview the user relentlessly about the planin 39 of 1048, across 12 files
  • order findings by severityin 29 of 1048
  • resolve each branch of the decision treein 28 of 1048, across 5 files
  • run a grilling sessionin 26 of 1048, across 5 files
  • update CONTEXT.md immediately when a term is resolvedin 26 of 1048, across 9 files
  • propose precise canonical terms for vague languagein 25 of 1048, across 6 files
  • create documentation files lazilyin 24 of 1048, across 5 files
  • use the domain-modeling skillin 22 of 1048, across 3 files

Said here and by no other author read

  • read the description and linked issue first
  • run non-trivial changes locally
  • review against author intent
  • label every comment blocking consider or nit
  • prioritize comments by correctness security design maintainability style
  • explain the reason for each comment

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.

Keep looking

Skills are one crate of 328,083. 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.