Back to list
mway

verification-before-completion

by mway

4🍴 0📅 Jan 22, 2026

SKILL.md


name: verification-before-completion description: Use when about to claim work is complete, fixed, or passing, or before committing/creating PRs; requires fresh verification evidence before any success claim. allowed-tools: ["shell", "read_file", "update_plan"] metadata: short-description: Evidence gate before completion claims

Verification Before Completion

Read these references:

  • ~/.config/agent/core/behavior.md - Evidence-based reasoning
  • ~/.config/agent/core/methodology.md - Verify phase discipline
  • ~/.config/agent/core/communication.md - Accurate status reporting

Overview

Claiming work is complete without verification is unacceptable. Evidence comes first.

Core principle: Evidence before claims, always.

Mode-Aware Verification

Verification requirements apply to ALL modes, but execution differs:

Normal mode:

  • Confirm before running verification commands
  • Show user the commands that will be run
  • Execute after confirmation

Autopilot mode:

  • Auto-run verification commands (Tier 3 operations)
  • Report progress: "Running tests..."
  • Show results automatically

Full-auto mode:

  • Auto-run verification commands
  • Log to audit trail
  • Only pause if verification fails repeatedly

Verification failure handling:

  • Normal: Report failure, ask for guidance
  • Autopilot: Attempt one auto-fix, then ask for guidance
  • Full-auto: Attempt up to 3 auto-fixes, then stop and report (see ~/.config/agent/domain/coding/guardrails.md recovery procedures)

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 the command that proves the claim
  2. Run the full command (fresh, complete)
  3. Read the full output and exit code
  4. Verify the output supports the claim
    • If NO: state actual status with evidence
    • If YES: state the claim with evidence
  5. Only then make the claim

Skipping any step is a failure to verify.

Common Failures

ClaimRequiresNot Sufficient
Tests passTest output shows 0 failuresPrevious run, "should pass"
Linter cleanLinter output shows 0 errorsPartial check
Build succeedsBuild command exit 0Linter passing
Bug fixedTest of original symptom passesCode changed
Requirements metChecklist verified"Looks done"

Red Flags — STOP

  • Using "should", "probably", "seems"
  • Expressing satisfaction before verification
  • About to commit/PR without verification
  • Trusting agent success reports without evidence
  • Relying on partial checks

Examples

Tests:

✅ Run tests → see 0 failures → "All tests pass"
❌ "Should pass now"

Build:

✅ Run build → exit 0 → "Build passes"
❌ "Linter passed"

Requirements:

✅ Re-read plan → verify checklist → "Requirements met"
❌ "Seems complete"

If verification fails or behavior is unclear, switch to systematic-debugging.

Common Verification Commands (Quick Reference)

Language/ToolingCommand
Gogo test ./...
Pythonpytest
Nodenpm test
Rustcargo test

When to Apply

Always before:

  • Any completion claim
  • Any positive status statement
  • Committing, PR creation, or moving to next task

Arguments

Target: ${ARGUMENTS}

Score

Total Score

40/100

Based on repository quality metrics

SKILL.md

SKILL.mdファイルが含まれている

+20
LICENSE

ライセンスが設定されている

0/10
説明文

100文字以上の説明がある

0/10
人気

GitHub Stars 100以上

0/15
最近の活動

3ヶ月以内に更新がある

0/10
フォーク

10回以上フォークされている

0/5
Issue管理

オープンIssueが50未満

+5
言語

プログラミング言語が設定されている

+5
タグ

1つ以上のタグが設定されている

0/5

Reviews

💬

Reviews coming soon