Verification Before Completion

---
name: verification-before-completion
description: Requires running verification commands and reading their actual output before any claim that work is complete, fixed, or passing. Use when about to say done, tests pass, or bug fixed; before committing, a PR, or trusting a subagent's report; or when asked to run the full quality gate (build, types, lint, tests, secrets, diff).
---

# Verification Before Completion

## Overview

**Core principle:** Evidence before claims, always.

**Violating the letter of this rule is violating the spirit of this rule.**

**Announce at start:** "Using opm:verification-before-completion before claiming this is done."

## The Iron Law

```
NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE
```

If you have not run the verification command in this message, you cannot claim it passes.

## The Gate Function

```
BEFORE claiming any status or expressing satisfaction:

1. IDENTIFY: What command proves this claim?
2. RUN: Execute the FULL command (fresh, complete)
3. READ: Full output, check exit code, count failures
4. VERIFY: Does output confirm the claim?
   - If NO: State actual status with evidence
   - If YES: State claim WITH evidence
5. ONLY THEN: Make the claim

Skip any step = lying, not verifying
```

## Quick Gate or Full Gate

- **Quick gate:** the Gate Function above, for any done claim.
- **Full gate:** before a PR, a release, or a phase close (or when asked for the full verification loop), read `references/full-loop.md` and run all six gates: build, types, lint, tests, secrets, diff review.

## Claim -> Required Evidence

| Claim | Requires | Not Sufficient |
|-------|----------|----------------|
| Tests pass | Test command output: 0 failures | Previous run, "should pass" |
| Linter clean | Linter output: 0 errors | Partial check, extrapolation |
| Type check clean | Type checker output: 0 errors | Editor shows no squiggles |
| Build succeeds | Build command: exit 0 | Linter passing, logs look good |
| Bug fixed | Test of original symptom: passes | Code changed, assumed fixed |
| Regression test works | Red-green cycle verified | Test passes once |
| Fast / smooth / battery-friendly | Measured on physical-device or prod-like, env stated | A simulator or unit run: report NOT JUDGEABLE |
| Subagent completed | `git diff` / `git log` shows the changes | Agent reports "success" |
| Requirements met | Line-by-line checklist against spec/plan | Tests passing |
| Plan task complete | Task's own test steps run, output read | Implementer said DONE |
| No silent failures | Error paths exercised or reviewed (`silent-failure-hunter` subagent for large diffs) | "I handled errors" |
| Safe to ship | `security-reviewer` subagent or `/security-review` on the diff | "Nothing sensitive touched" |

## Red Flags - STOP

- Using "should", "probably", "seems to"
- Expressing satisfaction before verification ("Great!", "Perfect!", "Done!")
- About to commit, push, or open a PR without verification
- Trusting a subagent's success report without checking the diff
- Relying on partial verification (one test file, one platform)
- Thinking "just this once"
- Tired and wanting the work to be over
- **ANY wording implying success without having run verification**

## Rationalization Prevention

| Excuse | Reality |
|--------|---------|
| "Should work now" | RUN the verification. |
| "I'm confident" | Confidence is not evidence. |
| "Just this once" | No exceptions. |
| "Linter passed" | Linter is not the compiler, and neither is the test suite. |
| "Agent said success" | Verify independently: read the diff, run the tests. |
| "I ran it earlier" | Earlier is before the last edit. Run it again. |
| "I'm tired" | Exhaustion is not an excuse. |
| "Partial check is enough" | Partial proves nothing about the rest. |
| "Different words so the rule doesn't apply" | Spirit over letter. |
| "The reviewer approved it" | Review is not execution. Run the command. |

## Key Patterns

**Tests:**
```
OK:  [Run test command] [See: 34/34 pass] "All tests pass"
BAD: "Should pass now" / "Looks correct"
```

**Regression tests (TDD red-green, see opm:tdd-workflow):**
```
OK:  Write -> Run (pass) -> Revert fix -> Run (MUST FAIL) -> Restore -> Run (pass)
BAD: "I've written a regression test" (without red-green verification)
```

**Build:**
```
OK:  [Run build] [See: exit 0] "Build passes"
BAD: "Linter passed" (linter does not check compilation)
```

**Requirements:**
```
OK:  Re-read plan/spec -> Create checklist -> Verify each item -> Report gaps or completion
BAD: "Tests pass, phase complete"
```

**Subagent delegation:**
```
OK:  Agent reports success -> Check git diff -> Run the task's tests -> Report actual state
BAD: Trust the agent's report
```

## When To Apply

**ALWAYS before:**
- ANY variation of a success or completion claim
- ANY expression of satisfaction
- ANY positive statement about the state of the work
- Committing, PR creation, task completion
- Moving to the next task in `opm:executing-plans`
- Delegating to, or accepting results from, subagents

**The rule applies to:**
- Exact phrases
- Paraphrases and synonyms
- Implications of success
- ANY communication suggesting completion or correctness

## Reporting Format

When the evidence is in, report it, not just the conclusion:

```
Verification:
- `npm test` -> 112 passed, 0 failed (exit 0)
- `npm run lint` -> 0 errors, 0 warnings
- `npm run build` -> exit 0
- Spec checklist: 7/7 requirements have a passing test
```

If any line fails, the work is not done. Say so, with the failing output, and keep going.

<!-- Adapted from obra/superpowers (MIT) -->