skills.rest
pgoell
pgoellCOMMUNITY
Explore all skills by this author →
Website𝕏@pagoellGitHub

verification-before-completion

Requires fresh command output as evidence before claiming work is complete.

Updated Mar 18, 2026
Search Tags:#code quality#testing#pull request#verification#evidence#ci#completion claims
One-click install
npx skills add https://github.com/pgoell/pgoell-claude-tools --skill verification-before-completion-pgoell
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: verification-before-completion
Source: https://github.com/pgoell/pgoell-claude-tools/tree/main/plugins/workbench/skills/verification-before-completion
Command: npx skills add https://github.com/pgoell/pgoell-claude-tools --skill verification-before-completion-pgoell

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI agents and developers often claim work is done, fixed, or passing based on stale output, prior-session results, or confidence rather than fresh evidence. This Skill enforces a verification gate so no completion claim is made without running the proving command after the final edit and reading its output. ## Core Features & Use Cases - Evidence Gate: Blocks completion claims until the proving command or check has run after the final relevant edit and its output, exit code, and failure count have been read. - Claim-to-Evidence Mapping: Defines required evidence for common claims such as tests passing, lint being clean, builds succeeding, bugs being fixed, and PRs being ready. - Red Flag Detection: Flags risky wording like "done", "fixed", "passes", "ready", and "safe to merge" as triggers to stop and verify before committing, pushing, or opening a PR. - Use Case: Before handing a bug fix back to a reviewer, run the regression test, confirm it passes, and report the exact command and result instead of stating the fix "should work". ## Quick Start Before claiming this task is complete, run the relevant test or build command and report the actual output as evidence.

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 code?▼

Identify the command that proves your claim, such as a test suite or build, run it after your final edit, and read the output and exit code. Only report completion if the fresh evidence actually supports the claim.

What evidence is required before claiming tests pass?▼

Claiming tests pass requires fresh test output with a passing exit code and zero failures, produced after the final relevant edit. Output from a prior session or from before the last change does not count as proof.

Can I use an agent's report as proof that work is done?▼

No, an agent report alone is not sufficient evidence. You must check the version control diff and run the relevant verification commands yourself before accepting or forwarding a completion claim.

What should I report when verification cannot be run?▼

State explicitly that verification was not run, explain the concrete reason, and describe any partial checks as partial only. Never soften the wording or imply completion without evidence.

Why is pre-final-edit test output not valid proof?▼

Code changed after the test run can invalidate earlier results, so only output produced after the final relevant edit proves the current state. Confidence or earlier green runs are not substitutes for fresh evidence.

Related Skills

Explore More Software Engineering Skills→
hardisgroupcomhardisgroupcom
368

test

Build, lint, and test code changes with yarn commands.

Official
Intermediate
All-The-VibesAll-The-Vibes
50

land

Commit changes, push to the remote, and open a pull request for handoff.

Official
Advanced
Kastalien-ResearchKastalien-Research

capture-learning

Capture and structure session learnings for agent memory updates.

Official
Intermediate
blackwell-systemsblackwell-systems
102

lsp-verify

Verify LSP diagnostics, build, and tests after code edits.

Community
Advanced