The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted local data to determine agent behavior.\n
Ingestion points: SKILL.md (Workflow steps 1 and 2) instructs the agent to run git diff --staged and git log to gather context.\n
Boundary markers: Absent; the output of git commands is ingested directly into the prompt context without delimiters or warnings to ignore embedded instructions.\n
Capability inventory: references/quick-reference.md lists git commit commands which are triggered as the final step of the workflow.\n
Sanitization: Absent; there is no validation or escaping of the content retrieved from the repository state.\n- [COMMAND_EXECUTION]: The skill facilitates the execution of local git commands as part of its core functionality.\n
Evidence: references/quick-reference.md provides bash commands for git diff, git log, git status, and git commit. These are standard tools for the skill's intended purpose of managing version control commits.

git-commit