The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local development commands such as git status, git diff, git add, and git commit. It also dynamically executes project-specific linting tools including npm run lint, ruff check, make lint, rubocop, and golangci-lint based on the detected project configuration.
[DATA_EXPOSURE]: The skill reads repository state and file differences (git diff) to summarize changes. It also checks for the existence of configuration files like package.json or pyproject.toml to determine the project type.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from git diff outputs. An attacker could potentially embed malicious instructions within code comments or documentation in a pull request to influence the generated commit message or the semantic analysis performed by the parallel agents.
Ingestion points: git diff output (SKILL.md)
Boundary markers: Absent. The prompts for semantic and breaking change analysis do not use specific delimiters or instructions to ignore embedded commands.
Capability inventory: Shell command execution (git, npm, ruff, etc.) and file writing via TodoWrite (SKILL.md).
Sanitization: None detected. The skill directly interpolates diff content into analysis prompts.

git-commit