enterprise-verify
Skill1 starsChanged 8 months agoMITView on GitHub
Evidence-based verification before any completion claim. No 'should work' or 'probably fine' — paste fresh test output or don't claim done. 7-check verification sequence. Use before committing or claiming work is complete.
1 other project uses a version of this, 1 of them edited.
Adopted by
Most recently taken up by
- anotherben/claude-harnessedited, first seen Nov 8, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 1
- anotherben/claude-harness★ 723% same
Wrote their own version · 0
None found.
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 repository11 other files it gives its agents.
- enterprise-brainstormSkill
- enterprise-buildSkill
- enterprise-compoundSkill
- enterprise-contractSkill
- enterprise-debugSkill
- enterprise-discoverSkill
- enterprise-forgeSkill
- enterprise-harnessSkill
- enterprise-planSkill
- enterprise-reviewSkill
- enterpriseSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 55
If overall is FAIL: stop, fix the failing checks, re-run verify.sh from scratch.
- Distinctive: line 116
When a check fails:
- Distinctive: line 119
If tests fail after a fix: you introduced a regression — revert and try again
- Distinctive: line 125
If you catch yourself reaching for these phrases, you haven't run verification yet:
- Very specific: line 139
Test suite takes too long?: Run verify.sh once with full suite.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Enterprise Verify — Evidence-Based Completion Verification
Why Evidence Matters
Without verification evidence, completion claims are beliefs — and beliefs ship bugs. "The code looks correct" means nothing when an import path is wrong, a test is silently skipped, or a debug console.log made it into the diff. Every check category below has caused a production incident at least once.
The distinction is simple: evidence is command output you can paste. Belief is anything else. This skill deals only in evidence.
Step 1: Run verify.sh
The verification script runs all 7 checks mechanically and produces structured JSON evidence. Run it — don't run the checks manually.
bash .claude/skills/enterprise-verify/scripts/verify.sh \
--base dev \
--contract <path-to-contract> \
--output .claude/enterprise-state/<slug>-verification-evidence.jsonFlags:
--base <branch>— base branch for diffs (default:dev)--contract <path>— contract file for postcondition extraction (optional)--skip-build— skip frontend build check (backend-only changes)--output <path>— where to write evidence JSON--project <path>— project root (auto-detected from cwd)
Step 2: Read the JSON Evidence
The script outputs a JSON file with this structure:
{
"checks": {
"test_suite": { "result": "PASS|FAIL", "passed": N, "failed": N },
"postcondition_trace": { "result": "MANUAL", "postconditions": "...", "test_names": "..." },
"regression": { "result": "PASS|FAIL", "new_failures": N },
"build": { "result": "PASS|FAIL|SKIP" },
"diff": { "result": "PASS", "files": [...] },
"imports": { "result": "PASS|FAIL", "unresolved": [...] },
"debug_artifacts": { "result": "PASS|FAIL", "findings": [...] }
},
"overall": "PASS|FAIL"
}If overall is FAIL: stop, fix the failing checks, re-run verify.sh from scratch.
Step 3: Complete the Two Manual Checks
The script automates 5 of 7 checks fully and collects data for the other 2. You must complete:
Postcondition Trace (Check 2)
Map each postcondition from the contract to a specific test name from the test_names field. Use the EXACT test description from the runner output — not a paraphrase. If a postcondition has no matching test: FAIL.
Diff Classification (Check 5)
The script lists changed files. Classify each as:
- REQUIRED — in the contract/plan
- ENABLING — needed to support a required change
- DRIFT — not related to this task →
git checkout -- path/to/file
Step 4: Write the Verification Report
===============================================
ENTERPRISE VERIFICATION REPORT
===============================================
Task: [what was done]
Date: [YYYY-MM-DD]
Branch: [branch name]
Evidence: [path to JSON file]
## Automated Checks (from verify.sh)
Check 1 — Test Suite: [PASS — N passed, 0 failed]
Check 3 — Regression Check: [PASS — no regressions]
Check 4 — Build Verification: [PASS / SKIPPED (backend only)]
Check 6 — Import Resolution: [PASS — all imports resolve]
Check 7 — Debug Artifacts: [PASS — none found]
## Manual Checks
Check 2 — Postcondition Trace: [PASS — N/N verified]
Check 5 — Final Diff: [PASS — N files, 0 drift]
────────────────────────────
OVERALL: [PASS / FAIL]When to Run
| Trigger | Action |
| About to say "it's done" | Run verify.sh first |
| About to commit | Run verify.sh first |
| About to create a PR | Run verify.sh first |
| "One more thing" was fixed | Re-run verify.sh from scratch |
Failure Recovery
When a check fails:
- Fix the issue
- Re-run verify.sh entirely (not just the failed check)
- If tests fail after a fix: you introduced a regression — revert and try again
Language That Signals Missing Evidence
If you catch yourself reaching for these phrases, you haven't run verification yet:
- "should work" / "probably fine" → run verify.sh, read the JSON
- "looks good" / "looks correct" → run verify.sh, report evidence
- "I'm confident" / "I believe" → confidence isn't evidence
Edge Cases
No formal contract? Derive postconditions from the task description. Run verify.sh without --contract — it still runs all other checks.
Only one file changed? Still run verify.sh. A one-line change can break tests or leave debug artifacts.
Test suite takes too long? Run verify.sh once with full suite. Don't substitute partial runs.
See references/verification-checks.md for the detailed command reference behind each check.
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.

