The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing various shell commands using git, gh, and sl to manage branches, commits, and pull requests. This includes operations like git reset, git commit --amend, and gh pr edit.
[DYNAMIC_EXECUTION]: The skill instructions direct the agent to build shell commands by interpolating commit titles into strings, such as git commit --only -m "<title>". While variables are wrapped in double quotes, there is a risk of command injection if commit titles contain malicious shell escape characters.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from local commit history, creating a surface for indirect prompt injection.
Ingestion points: The skill reads commit hashes, titles, and bodies using git log and sl log (SKILL.md, Step 1 and Sapling Step 1).
Boundary markers: No explicit boundary markers or safety instructions (like 'ignore embedded instructions') are used when processing these commit messages.
Capability inventory: The skill uses git, gh, and sl CLI tools and performs file writes to the local .git directory (SKILL.md, Step 1).
Sanitization: There is no mention of sanitizing or validating commit titles or descriptions before they are used in shell commands or PR properties.

submit-stack