| name | hands-on-verifier |
| description | Verify that generated hands-on code scripts run successfully and interface with local LLMs as expected. |
Hands-On Verifier Skill
This skill allows the agent to automatically verify the correctness and execution of the generated hands-on scripts (hands-on/main.py, hands-on/function_calling.py, hands-on/structured_output.py, hands-on/thinking_toggle.py).
When to use this skill
- Use this after generating or editing any code in the
hands-on/ directory to ensure that it has no syntax errors and runs successfully.
- Use this to verify that the generated code satisfies all the specific requirements from the rules files.
Verification Checklist
1. Code Quality & Syntax Check
- Verify that the target python file exists and is located in the
hands-on/ directory.
- Perform a syntax check (e.g., by running
python -m py_compile <file>).
2. Functional & Integration Execution
- Read the
.env configuration file in the hands-on/ folder to check which provider (LLM_PROVIDER) is selected.
- If the corresponding local LLM server is online, execute a dry run of the script with a test prompt.
- Capture the output and verify:
- Streaming works as expected.
- The model's reasoning/thinking process (e.g., inside
<think> tags) is successfully parsed and styled (or filtered out if in --no-think mode for thinking_toggle.py).
- Output is printed cleanly without tracebacks.
3. Verification Report
- Summarize the verification status of the code (e.g., syntax correctness, expected file locations, configuration matching).
- Report any issues found in the script logic or environment variables.