The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs shell command execution using standard development tools including git and the GitHub CLI (gh) in files references/commands/commit.md, references/commands/commit-push-pr.md, and references/commands/clean_gone.md. These commands are used for staging changes, committing code, managing branches, and interacting with remote repositories.
[PROMPT_INJECTION]: The skill's architecture is vulnerable to indirect prompt injection (Category 8) due to its processing of untrusted repository data.
Ingestion points: The components references/commands/commit.md and references/commands/commit-push-pr.md ingest the output of git status, git diff HEAD, and git log directly into the prompt context to facilitate the generation of commit messages.
Boundary markers: The prompt uses standard Markdown headers to distinguish system instructions from repository output, which lacks robust isolation to prevent the agent from potentially executing instructions found within code diffs.
Capability inventory: The agent has permissions to perform git commits, push code, and create pull requests, which could be misused if an injection successfully influences the agent's logic.
Sanitization: The skill does not implement any validation or sanitization of the git output before presenting it to the language model.

commit-commands