The Agent Skills Directory

[PROMPT_INJECTION]: The instructions include a specific directive to bypass user confirmation for terminal commands, which attempts to override standard agent safety guardrails.
Evidence: The phrase "no confirmation needed" in Step 5 of the workflow description in SKILL.md.
[COMMAND_EXECUTION]: The skill workflow requires the execution of multiple shell commands to perform its primary function.
Evidence: The skill instructs the agent to run git status, git diff, git add, and git commit in the user's terminal.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from local repository changes that could contain malicious instructions designed to manipulate the final commit message.
Ingestion points: git diff and git diff --cached output is analyzed by the agent as part of the workflow in SKILL.md.
Boundary markers: Absent. There are no instructions to the agent to treat content within the diff as data rather than instructions.
Capability inventory: The agent is instructed to execute git commit, allowing it to modify the repository history.
Sanitization: Absent. The agent is encouraged to construct messages based directly on content found in the diff.

conventional-commit