The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local git commands such as git diff and git log to extract information from the current repository. This is part of the core functionality but involves executing shell-level commands.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from the repository's history and current state. 1. Ingestion points: The rules/change-pack.md file specifies that content is read from git diff and git log to create a DIFF_CONTEXT. 2. Boundary markers: Absent; there are no instructions or delimiters provided to the agent to distinguish between valid code changes and potential malicious instructions embedded within the diff text. 3. Capability inventory: The agent uses extracted data to generate branch names, commit messages, and PR descriptions, which could be manipulated by adversarial content in the diff. 4. Sanitization: Absent; the skill summarizes the raw input into a structured context without filtering for potential prompt injection patterns.

change-pack