What are the 'Red Flags' this skill watches for?

It flags non-committal language like 'should', 'probably', 'seems to', or expressing satisfaction ('Great!', 'Done!') before running verification.

How does this improve software quality?

It prevents the common mistake of assuming a fix works. By requiring a fresh test run and exit code verification, it ensures only truly passing code is committed.

What is the core principle of this skill?

The core principle is 'Evidence before assertions.' Claude is prevented from claiming work is done until it has run the relevant verification commands and confirmed the output.

Can it be used for bug fixes?

Yes, it specifically includes a TDD Red-Green pattern where the developer must prove the test fails without the fix and passes with it.

Does it work when Claude delegates tasks to other agents?

Absolutely. It requires Claude to independently verify the changes made by other agents rather than blindly trusting their success reports.

Evidence-Based Completion Verification

Name: Evidence-Based Completion Verification
Author: LyQing63

byLyQing63

0•

Security & Testing

Enforces a rigorous evidence-first protocol that requires running verification commands before any task can be claimed as complete.

This skill instills a high-integrity workflow within Claude by mandating that every claim of success, completion, or bug-fix be supported by fresh, verifiable evidence. By blocking premature assertions like 'should work' or 'looks good,' it ensures that tests are run, builds are confirmed, and linters are checked before a task is finalized. This is essential for maintaining trust in automated workflows, preventing regression errors, and ensuring that code is production-ready before it reaches the commit or pull request stage.

Key Features

01Mandatory fresh verification command execution and output analysis

02Strict 'Evidence-Before-Claims' logic gate for all status updates

03Specific verification patterns for tests, builds, and linter checks

040 GitHub stars

05Anti-rationalization framework to prevent shortcuts and assumptions

06Independent verification protocols for agent-delegated tasks

Use Cases

01Verifying bug fixes through TDD red-green cycles to prove the original symptom is gone

02Ensuring all unit and integration tests pass before creating a pull request

03Confirming successful builds and linter compliance in complex CI/CD environments

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lyqing63/lyqing.opencode verification-before-completion

For use in Claude.ai and ChatGPT

Key Features

01Mandatory fresh verification command execution and output analysis

02Strict 'Evidence-Before-Claims' logic gate for all status updates

03Specific verification patterns for tests, builds, and linter checks

040 GitHub stars

05Anti-rationalization framework to prevent shortcuts and assumptions

06Independent verification protocols for agent-delegated tasks

Use Cases

01Verifying bug fixes through TDD red-green cycles to prove the original symptom is gone

02Ensuring all unit and integration tests pass before creating a pull request

03Confirming successful builds and linter compliance in complex CI/CD environments

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lyqing63/lyqing.opencode verification-before-completion

For use in Claude.ai and ChatGPT