The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill clones a project-specific repository from https://github.com/isaac-for-healthcare/i4h-workflows if the workflow directory is not found locally. This URL can also be overridden via the I4H_WORKFLOWS_REPO_URL environment variable.
[REMOTE_CODE_EXECUTION]: Following the repository checkout, the skill executes scripts (./run.sh and ./stop.sh) that are contained within the downloaded repository, which constitutes a remote code execution pattern.
[COMMAND_EXECUTION]: The skill uses several shell commands for repository management and discovery, including git clone, git rev-parse, and executing local bash scripts.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface through its ingestion of untrusted data.
Ingestion points: The skill reads DESIGN.md and the output of ./run.sh list from the cloned repository (SKILL.md).
Boundary markers: None identified; external content is processed directly to formulate responses.
Capability inventory: The skill has the capability to clone repositories, run shell scripts, and manage processes (SKILL.md).
Sanitization: No explicit sanitization or filtering of the external file content or script output is performed before incorporation into the agent's context.

i4h-workflow