The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from git diffs, logs, and file contents, creating a surface for indirect prompt injection.
Ingestion points: Git diffs, log history, and file contents are read in Step 1 and Step 5.
Boundary markers: The instructions do not explicitly define delimiters for the gathered diff content.
Capability inventory: The agent has the ability to execute git commands, including git add and git commit (Step 6).
Sanitization: A mandatory 'Risk Gate' (Step 3) is implemented to scan for secrets, tokens, and private keys before any staging or committing occurs.
[COMMAND_EXECUTION]: The skill uses standard git operations (status, diff, log, add, commit) as intended for its primary purpose. It incorporates security best practices by explicitly prohibiting the use of the --no-verify flag by default, ensuring that local repository hooks are executed.
[DATA_EXFILTRATION]: While the skill accesses sensitive file contents during the diff process, it lacks any network capabilities or external transmission commands, ensuring all data remains within the local environment.

commit