The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill processes untrusted external data which could contain instructions intended to manipulate the agent's output.
Ingestion points: Data enters the agent context through git diff, git status, and gh pr diff as specified in the Scope Modes and Workflow sections.
Boundary markers: The skill does not define specific delimiters or instructions to ignore natural language instructions found within the diff/code content.
Capability inventory: The skill is limited to read operations using git and gh CLI tools; it lacks write permissions, network exfiltration commands, or dynamic execution (eval/exec) capabilities.
Sanitization: No evidence of content filtering or sanitization of the retrieved diffs before they are passed to the LLM for summarization.
Command Execution (SAFE): The use of git and GitHub CLI (gh) is standard and necessary for the stated functionality of a change summarization tool. No high-risk command patterns (e.g., piping to shell, sudo) were detected.

summarize-changes