The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill dynamically generates Python test files (test/tests/test_{snake_name}.py) using a template in reference.md and then executes them using pytest via the Bash tool. While execution occurs inside a Docker container, the pattern of script generation and execution presents a risk if the generated content is influenced by malicious input.
[COMMAND_EXECUTION]: The skill executes docker compose commands to run the generated tests. This grants the agent the ability to manage Docker containers on the host system via the Bash tool.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface where untrusted data is processed and used to influence code execution.
Ingestion points: Ingests user-supplied strings via the AskUserQuestion tool in SKILL.md (Step 2).
Boundary markers: No markers or escaping are used in the Python template in reference.md to encapsulate user input, such as {PROMPT} and {JUDGE_QUESTION}.
Capability inventory: Utilizes the Write tool to create scripts and the Bash tool to execute them.
Sanitization: Lacks sanitization for inputs interpolated into the generated script, potentially allowing for code injection if a user or external source provides malicious Python syntax.

test-skill