The Agent Skills Directory

[Prompt Injection] (LOW): Indirect Prompt Injection surface. The skill ingests commit messages which are externally controlled data.
Ingestion points: Commit history retrieved via git log in SKILL.md.
Boundary markers: Absent; the prompt does not use delimiters to wrap the commit data.
Capability inventory: Minimal; the agent only performs summarization and formatting. No file-write or network capabilities are present.
Sanitization: None; the agent processes raw commit strings.
[Command Execution] (LOW): The skill executes git describe and git log. While these are shell commands, they are hardcoded and do not incorporate unvalidated user input into the command strings, making them low risk.

release-notes