The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection by processing untrusted repository metadata and file content.
Ingestion points: The skill analyzes repository state, staged/unstaged files, and diffs via git status (SKILL.md).
Boundary markers: There are no explicit markers or instructions to isolate the agent from potentially malicious instructions embedded in filenames or file content.
Capability inventory: The skill modifies the repository state via git commit and performs network operations via git push and pull request creation (SKILL.md).
Sanitization: No validation or sanitization of repository data is performed before it is used to generate commit messages or plan staging logic.
[COMMAND_EXECUTION]: The skill uses shell commands including git add, git branch, git commit, and git push to perform its intended functions. These are standard operations for Git-based workflows.

ship-pr