The Agent Skills Directory

[DYNAMIC_EXECUTION]: The implementation in scripts/opening_prs.py includes a mechanism to dynamically load the flowing library from a specific local path (/mnt/skills/user/flowing/scripts/flowing.py) if it is not already present in the Python environment. This is used for dependency resolution between sibling skills in the execution environment.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a conduit for user-supplied or agent-generated content to be pushed to GitHub. This inherent functionality presents an attack surface where malicious data could be processed and transmitted.
Ingestion points: The files, title, and body arguments in the open_pr function within scripts/opening_prs.py.
Boundary markers: The implementation does not include specific delimiters or instructions to ignore embedded commands in the PR content.
Capability inventory: The skill has the capability to write to the GitHub API, creating commits and pull requests.
Sanitization: There is no sanitization or escaping of the provided content before it is sent to the GitHub API.

opening-prs