The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill's execution_mode parameter includes an autopilot setting. This allows the agent to "Self-fix or delegate to subagents" if pre-commit hooks fail, which implies the generation and execution of code or system commands to resolve issues dynamically at runtime.
[COMMAND_EXECUTION]: The skill executes various git commands, including git add, git commit, and git log, to interact with the version control system.
[INDIRECT_PROMPT_INJECTION]: The workflow involves reading untrusted data from the local repository, creating a potential path for malicious instructions to influence agent behavior.
Ingestion points: The agent reads the .gitmessage file from the repository root and recent commit messages via git log to discover conventions.
Boundary markers: The instructions do not define boundary markers or explicit prompts to treat the content of .gitmessage as data only.
Capability inventory: The skill has access to the local file system and shell execution through git commands and the autopilot self-fix mechanism.
Sanitization: There is no evidence of sanitization or validation performed on the retrieved commit template or log history before they are used to guide message generation and execution.

vcs-git-commit