The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill implements a workflow (defined in SKILL.md steps 9 and 11) where the agent is directed to generate new Python scripts and tests based on user prompts and then execute them using python3 or a task runner (mise). This 'write-then-execute' pattern for agent-created code is a significant security risk, as it permits the execution of arbitrary logic generated at runtime.
[COMMAND_EXECUTION]: Multiple components of the skill (scripts/tests/cli.py, assets/starter-test.py, scripts/tests/test_scaffold_skill.py) utilize subprocess.run to execute Python scripts and unittest discovery. While these are used for internal validation and testing of the skill factory itself, they represent powerful execution capabilities within the agent's environment.
[INDIRECT_PROMPT_INJECTION]: The skill has a large attack surface as its primary function is to process user-provided prompts to drive the generation of code and instructions. A malicious prompt could influence the agent to generate code that performs unauthorized actions (such as data exfiltration or file modification), which would then be executed during the mandatory 'mise run ci' or 'doctor' verification phases.
[SAFE]: The skill references external resources from the author's official GitHub repository (github.com/srinitude/skills) for registry lookups and template fetching. These are identified as legitimate vendor-owned resources and do not contribute to verdict escalation.

skill-factory