The Agent Skills Directory

PROMPT_INJECTION (LOW): The skill is susceptible to indirect prompt injection because it processes untrusted code diffs and commit history.
Ingestion points: Analyzes output from git diff --staged and git log (SKILL.md, Instructions 1 and 5).
Boundary markers: Absent. There are no delimiters or instructions to treat analyzed code as untrusted data.
Capability inventory: The skill only generates text (commit messages) and does not have the authority to commit code, write files, or access the network.
Sanitization: Absent. The agent is not instructed to ignore malicious commands embedded in comments or code strings.
COMMAND_EXECUTION (LOW): The skill invokes local git commands.
Evidence: Executing git diff --staged and git log --oneline -10 to gather context.
Context: These are read-only operations used for their intended purpose in a developer-focused tool.

commit-helper