agentleFS
ChristopherAlphonseChristopherAlphonse/calphonse-skills/test-review/SKILL.md

test-review

Skill1 starsChanged yesterdayMITView on GitHub

Judge whether a diff is adequately tested, for code review. Runs focused sub-agents one after another (regression, behavior and edge cases, concurrency and race conditions, mutation) and reports findings with proof. Reads and runs, never edits the repo. Use for "review the tests", "are these tests good", "is this tested enough", "mutation test this", "test review", or as the test pass of a PR review.

19 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use test-plan, maintain-verification-skill and 6 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • mbwsimsSkill

    test-plan

    This skill should be used when the user asks to "plan tests", "what should I test", "what tests do I need", "test plan for", "testing strategy for", "test matrix for", "what cases should I cover", "help me plan tests", mentions "/test-plan", or wants to understand what tests are needed before writing any test code.

    Also in 3 projects that use this

  • jaydubya818Skill

    maintain-verification-skill

    Use when a project's verification driver or feature map may be stale.

    Also in 3 projects that use this

  • jaydubya818Skill

    create-verification-skill

    Use when a repository needs a repeatable driver that proves real UI, CLI, or service behavior.

    Also in 3 projects that use this

  • MicrosoftSkill

    blast-radius

    Assesses network dependency and shared-infrastructure exposure. Use when the user asks which services, customers, SLA policies, routes, or redundant links are affected by a network fault.

    Also in 3 projects that use this

  • asterkinSkill

    adr

    Create Architecture Decision Records (ADRs) following project standards. Guides template usage, process, and best practices for documenting architecturally significant decisions.

    Also in 3 projects that use this

  • AnthropicSkill

    architecture

    Create or evaluate an architecture decision record (ADR). Use when choosing between technologies (e.g., Kafka vs SQS), documenting a design decision with trade-offs and consequences, reviewing a system design proposal, or designing a new component from requirements and constraints.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 4 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 13 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Never edit, stage, or commit files in the user's checkout.

    line 13
  2. Distinctive:

    Do not install a tool or dependency without asking.

    line 15
  3. Very specific:

    Every finding is [ASSUMPTION], VERIFIED (read), or VERIFIED (ran).

    line 16
  4. Distinctive:

    Do not report style, naming, or test-framework preferences.

    line 17
  5. Very specific:

    Create the worktree outside the repo: git worktree add --detach <tmp>/test-review-<short-sha> HEAD.

    line 24
  6. Very specific:

    Create the handoff file .planning/reviews/test-review/<branch>.md in the target project.

    line 25
  7. Distinctive:

    Run one sub-agent per row, in this order, each only after the previous one has finished.

    line 29
  8. Very specific:

    Use the Agent tool with subagent_type: general-purpose, in the foreground, one call per row.

    line 42
  9. Very specific:

    Follow references/report.md.

    line 57

“Common” is measured: kept by at least 6 in 10 of the 13 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Test Review

Judge the tests that go with a change, not the change itself. The question is: if this diff were wrong, would the tests notice? Each focus is a separate sub-agent. They run in order, and each reads what the earlier ones found.

Guardrails

  • Scope is the diff against the base branch. Report only gaps in tests for the lines the diff adds or changes. Do not report missing tests for code the author did not touch.
  • Never edit, stage, or commit files in the user's checkout. All runs happen in a throwaway git worktree.
  • Throwaway tests are allowed inside the worktree to prove a gap. Never present them as the author's tests.
  • Do not install a tool or dependency without asking. If a tool is missing, say so and use the fallback in the focus reference.
  • Every finding is [ASSUMPTION], VERIFIED (read), or VERIFIED (ran). A VERIFIED (ran) finding quotes the command and its output. Omit a finding you cannot support at the first tier.
  • Do not report style, naming, or test-framework preferences.
  • This skill does not fix anything. A gap comes with a one-sentence description of the test that would close it.

Setup

  1. Resolve the base branch: git symbolic-ref --short refs/remotes/origin/HEAD, falling back to origin/main. If the branch is the base branch or the diff is empty, say "Nothing to review" and stop.
  2. Get the diff with git diff <base>...HEAD. If it cannot be obtained, stop and ask.
  3. Create the worktree outside the repo: git worktree add --detach <tmp>/test-review-<short-sha> HEAD. Remove it at the end.
  4. Create the handoff file .planning/reviews/test-review/<branch>.md in the target project. Create the directory if needed.

Pipeline

Run one sub-agent per row, in this order, each only after the previous one has finished. Never start two at once. Skip a row only when Scope says it does not apply, and record the skip and the reason in the handoff file.

#FocusQuestion it answersRead
1ScopeWhat changed, which test runner and tools exist, which focuses apply?references/scope.md
2RegressionDoes something that worked before now depend on a changed line without a test pinning it? Does a bug-fix test fail on the old code?references/regression.md
3BehaviorDo the tests assert what a caller can observe, including edge and failure cases?references/behavior.md
4ConcurrencyCan the changed code race, reorder, or deadlock, and does any test force it? Are the changed tests flaky?references/concurrency.md
5MutationIf the changed code were subtly wrong, would any test fail?references/mutation.md
6ReportWhat is the verdict?references/report.md

How to spawn each stage

Use the Agent tool with subagent_type: general-purpose, in the foreground, one call per row. Give every stage a standalone brief with:

  • the goal: "Judge whether this diff is adequately tested, for code review",
  • the base branch, the worktree path, and the handoff file path,
  • the focus reference to read and follow,
  • the rule that it reads the handoff file first and appends its own section, and does not rewrite earlier sections,
  • the guardrails above, repeated in full,
  • the report format: each finding has a severity, file:line, evidence tier, the gap, and the test that would close it.

Read the handoff file after each stage. If a stage's section is missing, or a finding cites a line that is not in the diff, send the stage back once with the problem named. After a second miss, record a gap and continue. A gap is not a pass.

You own the result. Do not paste a stage's output into the report without checking that its cited lines exist.

Output

Follow references/report.md. In short:

  • A verdict: APPROVE, REQUEST TESTS, or BLOCK.
  • Findings ordered by severity, each tied to a changed line, with its evidence tier and the test that would close it.
  • A table of stages with their status: ran, skipped (with reason), or gap.
  • What was not tested or not run, every time.

Durable output stays under .planning/reviews/test-review/ in the target project, never in this repo.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.