The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted repository data to generate commit messages.
Ingestion points: git diff HEAD and git log read data from the repository filesystem (SKILL.md).
Boundary markers: Absent; no explicit delimiters or instructions are used to separate repository content from the agent's instructions.
Capability inventory: git commit, git add.
Sanitization: Absent; the skill does not filter or escape the content gathered from the git history or diffs.
[SAFE]: The skill implements proactive security measures by explicitly instructing the agent to exclude sensitive files such as .env and credentials from the staging process.
[SAFE]: Includes branch protection logic that identifies the default branch (e.g., main or master) and requires explicit user confirmation before allowing a commit to those branches.
[COMMAND_EXECUTION]: The skill uses several git subcommands to manage version control. Access to these tools is correctly scoped and restricted within the YAML frontmatter configuration.

commit