Why is this better than just trusting the agent's report?

Agent reports can sometimes overlook side effects or partial failures; this skill requires independent verification of the actual system state and VCS diffs.

When does this skill activate?

It triggers whenever Claude is about to commit code, open a Pull Request, or communicate that a specific task or requirement has been satisfied.

Does it support Test Driven Development (TDD)?

Yes, it specifically enforces a red-green cycle for regression tests, requiring the test to fail when the fix is reverted to prove the test is valid.

How does this skill improve code quality?

It mandates that every claim of success—whether a fix, a build, or a test pass—is backed by fresh, logged evidence rather than assumptions or previous runs.

Strict Verification Before Completion

Name: Strict Verification Before Completion
Author: flosrn

byflosrn

0•

セキュリティとテスト

Enforces a rigorous evidence-based workflow that requires successful verification commands before any task is claimed as complete.

This skill implements a 'verify-first' protocol for Claude, preventing premature claims of success by requiring fresh, concrete evidence from test suites, build logs, and linter outputs. It establishes a mandatory 'Gate Function' that forces the execution of validation commands before any status is reported, eliminating common failures like 'it should work' or 'it seems fixed' in favor of raw terminal output and exit codes. This is essential for maintaining high-quality codebases and building trust in AI-human collaborations by ensuring no completion claim is made without verifiable proof.

主な機能

01Prevents the use of speculative language like 'should pass' or 'seems fixed'

020 GitHub stars

03Enforces a Gate Function requiring fresh command execution before status claims

04Requires 'Red-Green' cycles for regression tests to prove test validity

05Independently verifies agent-delegated work through VCS diffs and code reviews

06Validates builds, linters, and tests through direct exit code and output analysis

ユースケース

01Maintaining a high-trust development environment where AI assertions are always backed by logs

02Ensuring local builds and linters pass successfully before creating a Pull Request

03Preventing the submission of code changes that haven't been tested against original bug symptoms

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add flosrn/.claude verification-before-completion

For use in Claude.ai and ChatGPT

主な機能

01Prevents the use of speculative language like 'should pass' or 'seems fixed'

020 GitHub stars

03Enforces a Gate Function requiring fresh command execution before status claims

04Requires 'Red-Green' cycles for regression tests to prove test validity

05Independently verifies agent-delegated work through VCS diffs and code reviews

06Validates builds, linters, and tests through direct exit code and output analysis

ユースケース

01Maintaining a high-trust development environment where AI assertions are always backed by logs

02Ensuring local builds and linters pass successfully before creating a Pull Request

03Preventing the submission of code changes that haven't been tested against original bug symptoms