verification-before-completion

Enforces running verification commands before claiming work is complete or passing.

Updated Jul 22, 2026
One-click install
npx skills add https://github.com/jeremylightsmith/relay-config --skill verification-before-completion-jeremylightsmith
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: verification-before-completion
Source: https://github.com/jeremylightsmith/relay-config/tree/main/.claude/skills/verification-before-completion
Command: npx skills add https://github.com/jeremylightsmith/relay-config --skill verification-before-completion-jeremylightsmith

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI agents and developers often claim work is done, tests pass, or bugs are fixed based on assumptions, partial checks, or stale results. This Skill eliminates unverified completion claims by requiring fresh evidence before any status assertion. ## Core Features & Use Cases - Evidence Gate: Blocks any completion claim until the verification command (e.g., mix precommit) has been run fresh and its full output and exit status inspected. - Red-Green Regression Proof: Requires regression tests to be validated by reverting the fix, confirming failure, then restoring and confirming a pass. - Subagent Verification: Mandates checking git diff to independently confirm changes rather than trusting a subagent's success report. - Use Case: Before committing or opening a PR, the agent runs the full precommit gate, reads the output, counts failures, and only then reports the result with evidence. ## Quick Start Before claiming this task is done, run the full verification gate and report the actual output and exit status.

Frequently Asked Questions about verification-before-completion

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify work is complete before committing?▼

Run the project's full verification gate fresh, such as mix precommit, then read the complete output and check the exit status. Only claim completion after the output confirms zero failures, and state the claim together with that evidence.

How to prove a regression test actually works?▼

Use the red-green cycle: write the test, confirm it passes with the fix, revert the fix and confirm the test fails, then restore the fix and confirm it passes again. A test that only passes once proves nothing.

Can I trust a subagent's success report without checking?▼

No. Independently verify by inspecting git diff to confirm the changes actually exist and match the claimed work. A subagent's report of success is not evidence of completion.

Why is a passing linter not enough to claim code works?▼

A linter like Credo only checks style and static issues; it is not the compiler or the test suite. Partial checks prove nothing about runtime behavior, so the full gate including compile, tests, and audits must pass.

When should the verification gate run in a workflow?▼

Run it before any completion claim, expression of satisfaction, commit, push, pull request, or move to the next task. The verification must be fresh within the current message, not a previous run.