verification-before-completion

Enforces running verification commands and checking output before claiming work is complete.

Updated Jul 21, 2026
One-click install
npx skills add https://github.com/techfundoffice/cats-seo-aiagent-staging --skill verification-before-completion-techfundoffice
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: verification-before-completion
Source: https://github.com/techfundoffice/cats-seo-aiagent-staging/tree/main/.agents/skills/obra-verification-before-completion
Command: npx skills add https://github.com/techfundoffice/cats-seo-aiagent-staging --skill verification-before-completion-techfundoffice

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI agents and developers often claim tasks are done, tests pass, or bugs are fixed without actually running the verification commands, leading to broken code, lost trust, and wasted rework. This Skill eliminates false completion claims by requiring fresh evidence before any success statement. ## Core Features & Use Cases - Evidence-First Gate: Defines a mandatory gate function (identify, run, read, verify, then claim) that must be completed before any completion, commit, or PR claim. - Failure Pattern Catalog: Provides tables of common false claims, red-flag phrases like "should pass" or "probably works", and rationalization excuses with their corrections. - Use Case: Before committing a bug fix, the agent runs the full test suite, confirms zero failures in the output, and only then reports the fix as verified, instead of assuming the code change worked. ## Quick Start Ask the agent to verify the fix with fresh test output before claiming the task is complete.

Frequently Asked Questions about verification-before-completion

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

FAQPage Schema
How do I stop an AI agent from claiming tasks are done without testing?

Require the agent to run the actual verification command and quote its output before any completion claim. This Skill enforces a gate: identify the proving command, run it fresh, read the full output and exit code, then claim only what the evidence confirms.

What counts as valid verification before committing code?

Valid verification means running the full test suite, build, or linter in the current session and confirming zero failures from the actual output. Previous runs, partial checks, or assumptions like "should pass" do not qualify as evidence.

How do I verify a regression test actually catches the bug?

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 never fails proves nothing about the regression.

Can I trust a sub-agent's success report without checking?

No. Agent success reports are claims, not evidence. Independently verify by inspecting the version control diff and running the relevant verification commands before reporting the delegated work as complete.

When should verification run in an AI coding workflow?

Run verification before any success claim, expression of satisfaction, commit, pull request, task handoff, or move to the next task. The rule applies to exact phrases, paraphrases, and any implication that work is complete.