verify
Skill2 starsChanged 12 days agoMITView on GitHub
Review a diff against its ticket and the repo's standards, and report hard findings and judgment calls. Use when /ship or /build asks for a cold read, or the user asks to verify a diff.
189 other projects use a version of this.
Adopted by
Most recently taken up by
- SmartStudio/baize-bottheir own version, first seen Nov 15, 2026
- YuBan834/vrma-labtheir own version, first seen Nov 15, 2026
- Ju571nK/lodelinetheir own version, first seen Nov 15, 2026
- ericbust/namedroptheir own version, first seen Nov 15, 2026
- automagik-si/genietheir own version, first seen Nov 15, 2026
- todoist/todoist-clitheir own version, first seen Nov 15, 2026
Used with
Projects with it also use instinct-export, instinct-import and 10 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.
TheKREATORSSkillinstinct-export
Export instincts from project/global scope to a file
Also in 3 projects that use this
TheKREATORSSkillinstinct-import
Import instincts from file or URL into project/global scope
Also in 3 projects that use this
AviadPSkillrisk-assess
Assess the risk and blast radius of a proposed code change before implementing it. Evaluates CI impact, framework-wide vs localized scope, rollback path, and identifies blindspots. Use when user says "what's the risk", "is this safe", "assess the impact", "risk assess", "blast radius", or before applying framework-wide or CI-affecting changes. Do NOT use for code quality review (use code-quality) or plan validation (use design-validator).
Also in 3 projects that use this
diego-tobalinaSkillbuild-fix
Fix build and compilation errors systematically. Run when build fails.
Also in 6 projects that use this
AaravChadhaSkillsecure
Security review with a confidence gate - a finding exists only with a concrete exploit scenario and a high/medium/low confidence rating; everything else goes to a worth-hardening list, never inflated. Sweeps five surfaces: auth gates, secrets hygiene, injection and unsafe sinks, unsafe deserialization/crypto/transport, and LLM tool-use trust boundaries. Reports only, never fixes. Use when the user asks for a security review or to check vulnerabilities, secrets, or auth.
Also in 3 projects that use this
ArvorCoSkillspecify
Create feature specification from natural language description. Use when starting a new feature or creating a spec. Triggers on: create spec, specify feature, new feature spec.
Also in 10 projects that use this
- +1 more from TheKREATORS
Other versions189 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 189
- facebook/react★ 253ksame name
- react/react★ 252ksame name
- garrytan/gstack★ 142ksame name
- oven-sh/bun★ 97ksame name
- anthropics/financial-services-plugins★ 41ksame name
- anthropics/financial-services★ 41ksame name
- Yeachan-Heo/oh-my-claudecode★ 41ksame name
- Yeachan-Heo/oh-my-claude-sisyphus★ 41ksame name
- anthropics/claude-quickstarts★ 19ksame name
- NVIDIA-NeMo/NeMo★ 19ksame name
179 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository22 other files it gives its agents.
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 167 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 21
Use the diff and the criteria that the caller gave.
- Distinctive: line 39
Read the diff once for each heading.
- Very specific: line 56
When the diff adds a dependency, compare its engines with the versions in the CI workflow and the deploy target.
“Common” is measured: kept by at least 6 in 10 of the 167 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.
Files
Verify
/verify is the adversarial review of one diff. It reports and edits nothing. The caller fixes the findings.
A finding is hard or a judgment call. A hard finding blocks the ticket. A judgment call goes to the user.
1. Pin the diff
Use the diff and the criteria that the caller gave. When you run on your own:
- Issue: the argument, or the number in the branch name.
- Base: the ref. With no ref, use
git merge-base origin/<trunk> HEAD, because a local trunk can be ahead of the branch base. - Diff:
git diff <base>. It includes the work that is not committed yet.
Read the issue with its criteria, with the Read command of docs/agents/tracker.md, or as NO-SETUP.md in the folder of the setup-skills skill says, and the spec with the testing decisions, when there is one. Run git diff --stat <base> and git status --short. Read each untracked file: git diff does not show it.
Completion: the base resolves, the diff is not empty, you have the criteria when an issue exists, and you know each untracked file.
2. Run the checks
Find the test, typecheck, lint and build commands in package.json, the README or AGENTS.md. Run the typecheck, the lint and the build. Run the test files that the diff adds or changes, and each test file that imports a module that the diff changes. Run the full suite only when the caller asks for it, or when you run on your own. Trust no earlier report.
Completion: each check passed or failed, each run has finished, and you have the output of each failure.
3. Review
Read the diff once for each heading. Keep the two headings apart, so that one does not hide the other.
Spec: does the diff do what the ticket asks?
- Each criterion has code and a test that proves it.
- Each test is at a seam from the spec's testing decisions. With no spec, each test is at a seam that the existing tests of the area use.
- Each expected value comes from a literal, a worked example or the spec, not from the code's own logic.
- The result can be checked from outside: through the UI, an API or a command. A ticket that changes one layer only, such as storage with no UI, is not a vertical slice.
- Behavior that the ticket did not ask for is a judgment call.
For refactor work, the behavior must stay the same. Replace the first item and the vertical slice item with these: no expected value in an existing test changed, each test change follows a rename or a move, and each criterion is met in the structure of the code.
Standards: does the diff follow the rules of the repo?
- Find the written rules:
AGENTS.md,CLAUDE.md,CONTRIBUTING.md,CODING_STANDARDS.mdand the ADRs for the area. - Then read
SMELLS.mdin the folder of this skill with the Read tool, and look for its smells. A written rule wins over a smell. - Skip what a linter or a formatter already enforces.
- When the diff adds a dependency, compare its
engineswith the versions in the CI workflow and the deploy target.
Completion: each criterion, each test and each written rule is checked, and you read each changed or new test file.
4. Sort
A finding is hard when it is one of these:
- A check from step 2 fails.
- A criterion has no code, or no test proves it. A criterion that no agreed seam can test needs only the code.
- A test is at a seam that the spec did not agree, or it is tautological.
- The diff breaks a written rule of the repo.
- A new dependency needs a runtime version that CI or the deploy target does not have.
- The diff causes a defect that a user can see or that stores or sends wrong data, such as a false text, a lost value or a stale screen.
- The work is not a vertical slice. This does not apply to refactor work.
- Refactor work changed an expected value in an existing test.
Defects that a user can see were filed as judgment calls in 7 sessions of one retro. So before you file a judgment call, answer the user test for it. When any answer is yes, the finding is hard:
- Can a user see or hear it, in any state of the story: empty, at a limit, after an error, after a stop, pause or retry?
- Does it store, show, export or send a wrong value, such as a wrong currency, total or count?
- Does it lose what the user entered or chose, such as a filter, a toggle or an edit?
- Does a screen fail with no message, or a dialog take no focus or have no way to close?
- Does it write a secret, a token or personal data to a log?
To answer, walk each story of the criteria through its edge states in the code: empty, one, at the limit, error, and an action in the middle of another one. A finding that only one edge state shows is still hard.
Everything else is a judgment call. Keep the five that matter most.
Completion: each finding is hard or a judgment call, each judgment call passed the user test, and you have five judgment calls or fewer.
Report
## Hard
- <file>:<line> · <rule or criterion> · <what is wrong>
## Judgment calls
- <file>:<line> · <smell> · <fix in a few words> · user effect: none
and <n> more"user effect: none" records that the finding passed the user test of step 4. When you cannot write it, the finding is hard.
Print this block in each run, before the caller commits. Write "None" under a heading that has no findings. Write each finding on one line. Leave out the "and <n> more" line when you kept every judgment call.
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
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.

