verification-before-completion

name	verification-before-completion
description	Use before claiming work complete, fixed, or passing — and before committing or opening PRs. Run verification commands and confirm output first.
current_aal	2
target_aal	3

Verification Before Completion

Overview

Claiming work is complete without verification is dishonesty, not efficiency.

Core principle: Evidence before claims, always.

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

The Iron Law

NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE

If you haven't 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

Common Failures

Claim	Requires	Not Sufficient
Tests pass	Test command output: 0 failures	Previous run, "should pass"
Linter clean	Linter output: 0 errors	Partial check, extrapolation
Build succeeds	Build command: exit 0	Linter passing, logs look good
Bug fixed	Test original symptom: passes	Code changed, assumed fixed
Regression test works	Red-green cycle verified	Test passes once
Agent completed	VCS diff shows changes	Agent reports "success"
Requirements met	Line-by-line checklist	Tests passing

Red Flags - STOP

• Using "should", "probably", "seems to"
• Expressing satisfaction before verification ("Great!", "Perfect!", "Done!", etc.)
• About to commit/push/PR without verification
• Trusting agent success reports
• Relying on partial verification
• Thinking "just this once"
• Tired and wanting work over
• ANY wording implying success without having run verification

Rationalization Prevention

Excuse	Reality
"Should work now"	RUN the verification
"I'm confident"	Confidence ≠ evidence
"Just this once"	No exceptions
"Linter passed"	Linter ≠ compiler
"Agent said success"	Verify independently
"I'm tired"	Exhaustion ≠ excuse
"Partial check is enough"	Partial proves nothing
"Different words so rule doesn't apply"	Spirit over letter

Key Patterns

Tests:

✅ [Run test command] [See: 34/34 pass] "All tests pass"
❌ "Should pass now" / "Looks correct"

Regression tests (TDD Red-Green):

✅ Write → Run (pass) → Revert fix → Run (MUST FAIL) → Restore → Run (pass)
❌ "I've written a regression test" (without red-green verification)

Build:

✅ [Run build] [See: exit 0] "Build passes"
❌ "Linter passed" (linter doesn't check compilation)

Requirements:

✅ Re-read plan → Create checklist → Verify each → Report gaps or completion
❌ "Tests pass, phase complete"

Agent delegation:

✅ Agent reports success → Check VCS diff → Verify changes → Report actual state
❌ Trust agent report

Prototype-Patch Verification Gate

When a fix monkey-patches a shared runtime object (a class prototype, a module export, an interpreter-level hook), test-runner evidence is structurally insufficient: test runners commonly deduplicate module instances into one graph, while the production runtime may load two or more independent copies of the same library (e.g. dual-build packages resolved differently by different consumers). The patch then works in every test and never fires in production.

Gate: before claiming a prototype/module-patch fix complete, run a probe in the production-like runtime (container, deployed process — not the test runner) that asserts the patched object is the SAME instance every consumer uses. If the library ships multiple builds, resolve it through each loading mechanism the runtime uses and compare identities:

✅ [Probe in target runtime] [See: patched object identity === consumer's object identity] "Patch covers production"
❌ "All specs green" (test runner deduped the module graph; production splits it)

If identities differ, the fix must patch every reachable instance, and the probe becomes part of the verification suite.

Why This Matters

From 24 failure memories:

• your human partner said "I don't believe you" - trust broken
• Undefined functions shipped - would crash
• Missing requirements shipped - incomplete features
• Time wasted on false completion → redirect → rework
• Violates: "Honesty is a core value. If you lie, you'll be replaced."

When To Apply

ALWAYS before:

• ANY variation of success/completion claims
• ANY expression of satisfaction
• ANY positive statement about work state
• Committing, PR creation, task completion
• Moving to next task
• Delegating to agents

Rule applies to:

• Exact phrases
• Paraphrases and synonyms
• Implications of success
• ANY communication suggesting completion/correctness

The Bottom Line

No shortcuts for verification.

Run the command. Read the output. THEN claim the result.

This is non-negotiable.

Shared Working Tree — Verify the Committed Blob, Not the Tree

In a shared git working tree (a framework or workspace repo touched by multiple parallel agent sessions), another session may git checkout a different branch in the same tree after you committed your work. A direct grep / cat / Read of the working tree then measures the WRONG branch — you can read a sibling task's pre-fix state and raise a false failure against your own already-correct commit.

Rule: before verifying a task's output, probe the tree's current branch:

git -C <repo> rev-parse --abbrev-ref HEAD

If it differs from your task branch, read your committed state directly from your branch instead of the working tree:

# noshellcheck-extract
git -C <repo> cat-file blob <task-branch>:<path>   # or: git show <task-branch>:<path>

This applies at every verification gate (self-review, QA, compliance, archive). The working tree is shared state; your branch is the source of truth for what you shipped.

Shared Working Tree — Never Switch HEAD Out From Under a Sibling

The rule above covers the victim side: another session's checkout yanked your tree, so verify against your committed blob. This rule covers the actor side: do not be the session that does the yanking.

In a shared working tree, git checkout <branch> / git switch changes what every file on disk resolves to — for every process reading that tree, not just your own. If a sibling agent is mid-edit (uncommitted hunks, a long-running test run, a partially-written file), switching HEAD silently rewrites their in-progress files out from under them or drops their edits into the wrong branch's history when they eventually save.

Rule: in a shared working tree, never git checkout / git switch to change the tree's HEAD. Use an isolated worktree instead:

# noshellcheck-extract
git -C <repo> worktree add <path> <branch-or-ref>

This gets you an independent HEAD and working directory without touching the shared tree a sibling may depend on. See skills/using-git-worktrees/SKILL.md for the full isolated-workspace setup procedure.

Exception: a workspace repo with a single canonical branch (e.g. main only, no feature branches ever checked out) has no sibling-HEAD hazard — there is nothing to switch away from. The hazard is specific to repos where different sessions work on different branches/tasks in the same tree.