The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various shell commands including git status, git diff, git ls-files, git add, and git commit through a sub-agent to manage the repository state.
[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to instructions embedded in the repository's files or diffs that are processed at runtime.
Ingestion points: The sub-agent reads git diff output and the full content of all untracked files discovered via git ls-files (SKILL.md).
Boundary markers: There are no explicit boundary markers or "ignore instructions" prompts provided to the sub-agent to prevent it from obeying instructions found within the repository data.
Capability inventory: The sub-agent is authorized to execute shell commands, read local files, and modify the git history.
Sanitization: No sanitization, filtering, or validation is performed on the data ingested from the repository files before it is processed by the agent.

micro-commit