The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to interact with the repository, specifically git diff --cached --name-only and git diff --cached, and performs commits using HEREDOC syntax. These operations are within the scope of its primary purpose.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon data that may be controlled by an external attacker, creating a vulnerability surface.
Ingestion points: The skill reads staged diffs via git diff and incorporates "extra invocation text" and "session history findings" (motivation/rationale) into its logic.
Boundary markers: The instructions do not specify any delimiters or safety markers to differentiate between instructions and the data being processed from the diffs or history.
Capability inventory: The skill has the ability to read files, modify the README.md file, and execute git commit commands.
Sanitization: There is no evidence of sanitization, validation, or escaping of the ingested data before it is interpolated into commit messages or used to determine documentation updates.

commit-staged