Skill Tester
Purpose
Use this skill to make a skill run observable. The output is a local debug report that helps the user inspect what happened during a concrete test of another skill.
Do not turn this into an evaluation framework. Do not assign formal pass/fail status unless the user explicitly supplied pass/fail criteria.
Workflow
Identify the target skill and test request.
- Record the target skill name and path if known.
- Record the user's exact test request or a concise quoted paraphrase if the request is long.
- If the target skill or task is ambiguous, ask the minimum question needed before running the test.
Create a local report file before or at the start of the run.
- Prefer a project-local
.skill-test-runs/ directory under the current workspace.
- Create the directory if it does not exist.
- Use a conspicuous filename:
SKILL_TEST_REPORT_<skill-name>_<YYYYMMDD-HHMMSS>.md
- If the file already exists, append
-2, -3, and so on until the path is unused.
- Treat the report as local debugging output. Do not place it in tracked documentation unless the user explicitly asks.
Run the target skill normally on the concrete request.
- Follow the target skill's own instructions.
- Keep a running record of visible actions: files read, commands run, files edited, generated artifacts, tool failures, and user-facing decisions.
- Do not include hidden chain-of-thought or private reasoning. Summarize decisions and observations at a practical level.
Write the report.
- Include enough detail for the user to reproduce or debug the run.
- Redact tokens, credentials, cookies, private account identifiers, and other secrets if they appear in command output or files.
- Keep local/private machine paths only in the local report when needed for debugging; do not copy those details into public-facing repository docs.
Final response.
- Return the report path.
- Briefly state what was tested, what the target skill actually did, and whether any obvious blockers or uncertainties were observed.
Report Template
Use this shape unless the user requests a different format:
# Skill Test Report: <skill-name>
## Brief
- Target skill: `<skill-name>`
- Target path: `<path or unknown>`
- Test request: <user request>
- Run time: <local timestamp>
- Result summary: <what happened in 2-5 bullets>
## Visible Process
1. <step actually taken>
2. <step actually taken>
## Files Read
- `<path>` - <why it was read>
## Commands Run
```text
<command>
```
Outcome: <exit code and concise result>
## Files Changed Or Created
- `<path>` - <what changed>
## Outputs And Artifacts
- `<path or artifact>` - <what it contains>
## Errors, Ambiguities, And Blockers
- <issue or "None observed">
## Behavior Notes For User Debugging
- <what the target skill seemed to handle well>
- <where the target skill was unclear, overbroad, under-specified, or hard to follow>
Boundaries
- Do not claim the target skill is public-ready, safe, correct, or complete from this report alone.
- Do not automatically edit the target skill unless the user explicitly asks for fixes after the test.
- Do not create broad test matrices unless the user asks for a more formal test plan.
- Do not write hidden reasoning, system prompts, or private deliberation into the report.
1---2name: skill-tester3description: Record observable skill test runs for debugging. Use when the user wants to test, dry-run, debug, inspect, or compare how a Codex skill behaves on a concrete request; creates a local non-overwriting Markdown report with the visible process, files read, commands run, changed artifacts, errors, and a brief behavior summary. Does not certify skill quality or replace user judgment.4---56# Skill Tester78## Purpose910Use this skill to make a skill run observable. The output is a local debug report that helps the user inspect what happened during a concrete test of another skill.1112Do not turn this into an evaluation framework. Do not assign formal pass/fail status unless the user explicitly supplied pass/fail criteria.1314## Workflow15161. Identify the target skill and test request.17 - Record the target skill name and path if known.18 - Record the user's exact test request or a concise quoted paraphrase if the request is long.19 - If the target skill or task is ambiguous, ask the minimum question needed before running the test.20212. Create a local report file before or at the start of the run.22 - Prefer a project-local `.skill-test-runs/` directory under the current workspace.23 - Create the directory if it does not exist.24 - Use a conspicuous filename:2526```text27SKILL_TEST_REPORT_<skill-name>_<YYYYMMDD-HHMMSS>.md28```2930 - If the file already exists, append `-2`, `-3`, and so on until the path is unused.31 - Treat the report as local debugging output. Do not place it in tracked documentation unless the user explicitly asks.32333. Run the target skill normally on the concrete request.34 - Follow the target skill's own instructions.35 - Keep a running record of visible actions: files read, commands run, files edited, generated artifacts, tool failures, and user-facing decisions.36 - Do not include hidden chain-of-thought or private reasoning. Summarize decisions and observations at a practical level.37384. Write the report.39 - Include enough detail for the user to reproduce or debug the run.40 - Redact tokens, credentials, cookies, private account identifiers, and other secrets if they appear in command output or files.41 - Keep local/private machine paths only in the local report when needed for debugging; do not copy those details into public-facing repository docs.42435. Final response.44 - Return the report path.45 - Briefly state what was tested, what the target skill actually did, and whether any obvious blockers or uncertainties were observed.4647## Report Template4849Use this shape unless the user requests a different format:5051````markdown52# Skill Test Report: <skill-name>5354## Brief5556- Target skill: `<skill-name>`57- Target path: `<path or unknown>`58- Test request: <user request>59- Run time: <local timestamp>60- Result summary: <what happened in 2-5 bullets>6162## Visible Process63641. <step actually taken>652. <step actually taken>6667## Files Read6869- `<path>` - <why it was read>7071## Commands Run7273```text74<command>75```7677Outcome: <exit code and concise result>7879## Files Changed Or Created8081- `<path>` - <what changed>8283## Outputs And Artifacts8485- `<path or artifact>` - <what it contains>8687## Errors, Ambiguities, And Blockers8889- <issue or "None observed">9091## Behavior Notes For User Debugging9293- <what the target skill seemed to handle well>94- <where the target skill was unclear, overbroad, under-specified, or hard to follow>95````9697## Boundaries9899- Do not claim the target skill is public-ready, safe, correct, or complete from this report alone.100- Do not automatically edit the target skill unless the user explicitly asks for fixes after the test.101- Do not create broad test matrices unless the user asks for a more formal test plan.102- Do not write hidden reasoning, system prompts, or private deliberation into the report.
shenmuxing