The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external skill documents and task descriptions, which inherently creates a surface for indirect prompt injection. This risk is mitigated by the implementation of clear boundary markers in prompts and the use of a restricted temporary workspace that only copies an allowlist of necessary support files, preventing access to the broader file system during validation.
[COMMAND_EXECUTION]: The unit test file tests/test_run_post_merge_task.py uses the Python exec() function. This is used specifically to verify the logic of Python functions extracted from heredocs within the project's own shell scripts during testing. This dynamic execution is limited to verifying local source code and does not process external or untrusted input.

skillopt