Verify
Pre-PR gate: runs your full build → lint → test → secrets scan loop and produces a PASS / CLOSE / FAIL verdict with a prioritized fix list. Single agent. No auto-fix.
How It Works
- TRIGGER: Before you push — Pre-PR gate — diagnoses, never auto-fixes
- GATE: Detect stack & run build — Node · Python · Go · Rust — captures exit code (✕ fail → Build broke → stop & report FAIL immediately)
- STEP 1: Run linter & test suite — Lint errors + failing tests block; warnings + low coverage do not
- GATE: Scan diff for secrets — API keys · tokens · private keys (AWS · OpenAI · GitHub · Slack) (✕ fail → Any match → automatic FAIL, no exceptions)
- OUTPUT: Gate verdict — PASS / CLOSE / FAIL + numbered fix list for every blocker
- Before you push: Pre-PR gate — diagnoses, never auto-fixes
- Detect stack & run build: Node · Python · Go · Rust — captures exit code
- Run linter & test suite: Lint errors + failing tests block; warnings + low coverage do not
- Scan diff for secrets: API keys · tokens · private keys (AWS · OpenAI · GitHub · Slack)
- Gate verdict: PASS / CLOSE / FAIL + numbered fix list for every blocker
Invocation Triggers
/verifyverifypre-PR checkbefore I pushis my code readyUse Cases
- Run a full build + lint + test + secrets check before every push
- Get a single PASS / CLOSE / FAIL verdict with a numbered fix list
- Catch secrets in your diff before they enter git history
The Problem
You push. CI fails. Your team sees it before you do. Not because you missed something obvious — because you didn't have a fast way to check everything in one pass before you hit push. You ran the tests, maybe. You forgot lint. You definitely didn't scan for secrets. CI is not a feedback loop. It's a public scoreboard. Verify is the private rehearsal you run before the curtain goes up.
What It Does
- 1Detect stack and run build
Auto-detects Node, Python, Go, or Rust. Runs your build command and captures the exit code. If the build fails, the skill stops here — there is no point linting a broken codebase.
- 2Run your linter
Detects ESLint, Ruff, Flake8, golangci-lint, or Clippy. Errors block the verdict; warnings appear in the report but do not block a push.
- 3Run your test suite with coverage
Runs your existing test command with coverage flags. Any failing test is a blocker. Coverage below 80% is a warning, not a blocker.
- 4Scan for secrets
Pattern-matches against your git diff for API keys, tokens, private keys, and common credential formats (AWS, OpenAI, GitHub, Slack). Any match is an automatic FAIL — no exceptions.
- 5Produce the gate verdict
PASS, CLOSE, or FAIL with a numbered fix list for every blocker and a warnings section for non-blocking issues. One report, one verdict, nothing to interpret.
What You Get / What It Doesn't Do
- A single PASS / CLOSE / FAIL verdict before you push
- Build status with full error output if it failed
- Lint error count and top locations — errors vs warnings separated
- Test results with failing test names and current coverage %
- Secrets scan across your git diff with file:line references
- A numbered fix list for every blocker
- Fix build errors, lint violations, or failing tests
- Open pull requests or interact with GitHub
- Block your push automatically — it reports, you decide
- Replace CI — it is the check you run before CI sees your code
- Scan your full git history — scans the current diff only
Tips
This is a pre-push tool. By the time CI fails, your team has already seen it. Run verify locally first.
A secrets match means a credential is about to enter git history. Rotate it before you push — not after. History rewrites are painful.
CLOSE means you can push, but look at the warnings first. Lint warnings and coverage gaps that get ignored today become PRD-blocking debt tomorrow.
Get the Skill
Verify
The full SKILL.md — copy it into ~/.claude/skills/ and trigger it by name.
Commonly Used With
New skills ship here first
That one's yours — keep it. The library grows every week. Get each new production skill the day it drops, free, and follow the build along the way.
Get every new skill free30 production skills in the library — ready to install.