The Agent Skills Directory

[COMMAND_EXECUTION]: The skill automates version control by executing shell commands like git add and git commit.
[PROMPT_INJECTION]: In Step 5, the skill explicitly tells the agent to "proceed immediately without asking for confirmation". This behavior suppresses user oversight, which is a security concern when the agent performs write operations like committing code.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8). It analyzes untrusted content from git diff to generate commit messages and select files for staging without sanitization or boundary markers. 1. Ingestion points: The agent reads the output of git status and git diff in SKILL.md. 2. Boundary markers: None; untrusted diff data is processed directly. 3. Capability inventory: The skill can stage files (git add) and record changes (git commit) as seen in SKILL.md. 4. Sanitization: There is no logic to escape or validate the content of the changes being analyzed.

oma-commit