The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local repository in the form of code changes and diffs, creating a vulnerability surface where instructions embedded in the code could override the skill's logic.
Ingestion points: The skill instructions in SKILL.md direct the agent to run git status --short and git diff --cached to inspect changes.
Boundary markers: The instructions do not specify any delimiters or warnings to ignore instructions that might be present within the diff content (e.g., in code comments).
Capability inventory: The skill possesses the ability to execute git commit to modify the repository history and uses shell commands to interact with the file system.
Sanitization: There is no evidence of sanitization or validation of the diff content before it is processed by the agent to draft commit messages.
[COMMAND_EXECUTION]: The skill relies on the execution of local Git commands to perform its core functions.
The workflow involves running git status, git diff, and git commit via the command line interface.
While these are the primary intended functions of the skill, they represent an active command execution capability within the environment.

write-git-commits