How to use the check-tests agent skill: test evidence checks

What is an agent skill and what does check-tests make the agent do?

An agent skill is a folder containing a SKILL.md instruction file that an AI agent can load when a task matches it. The check-tests skill is named check-tests and is described as running the local test evidence check.

According to its instructions, the agent should keep source paths in its result and stop when evidence is missing. In other words, the agent reports what it found with the file paths attached, and it does not carry on as if the tests passed when no evidence exists.

• Looks for test evidence locally as part of the task.
• Includes source file paths in the reported result.
• Stops instead of continuing when evidence is missing.
• Defines no specific commands, frameworks or CI integration.

How do you install the check-tests skill?

The skill ships as a ZIP download from the source page. Unzip it into one of two locations, depending on how you want to use it.

For a personal skill available across your projects, unzip it into ~/.claude/skills/. For a skill that only applies to one project, unzip it into the project's .claude/skills/ folder.

If you use the Claude apps instead, you can upload the skill under Settings → Capabilities → Skills.

• Personal install: ~/.claude/skills/
• Project install: .claude/skills/ inside your project
• Claude apps: Settings → Capabilities → Skills, then upload

How do you trigger the check-tests skill with natural requests?

You do not need a special command. Describe the task in ordinary language so the agent recognises it as a test evidence check. The skill's description is about running the local test evidence check, so requests that mention checking tests, confirming changes are tested, or verifying test evidence fit its purpose.

Be explicit about the outcome you want. Asking the agent to list the source paths behind its conclusions reinforces the skill's requirement to keep those paths in the result.

• "Run the local test evidence check on these changes."
• "Confirm this change is backed by tests and list the source paths you used."
• "Check for test evidence before we move on, and stop if there isn't any."

What does a result look like?

The source does not define an output format, so the exact layout depends on the agent. What the skill does require is that the result keeps source paths and that the agent stops when evidence is missing.

Treat a stop as a useful answer. It means the check could not find the evidence it needed, and you should add or run tests before relying on the work.

• A result that cites source file paths is the expected outcome.
• A stop with no evidence found means the work is not confirmed as tested.

What does check-tests need, and what are its safety notes?

The skill lists no requirements. It runs no scripts and uses no network access, according to its listing. Its behaviour therefore depends on what the agent already knows about running tests in your project.

Because the skill defines no commands or frameworks, the agent must work out how your project's tests run on its own. If your project needs a particular test tool, state it in your request or in your project's own instructions so the agent does not guess. The skill also does not fetch or verify results from a CI service, so it only describes a local check.

• No listed requirements.
• Runs scripts: no.
• Uses network: no.
• Does not check CI service results.
• Set the test command in your request if your project needs a specific tool.

Tips for getting reliable results

Give the agent the context it lacks. Since the skill is only a short instruction file, a clear description of your test setup will do more than any extra wording in the request.

Review the source paths the agent returns. The value of this skill is traceability, so check that the cited files are the ones you expect.

• Name your test tool and how it is normally run.
• Ask for source paths in every result.
• Read the stop message as a signal to add tests, not as a glitch.