The Agent Skills Directory

Indirect Prompt Injection Surface: The skill is designed to ingest and analyze untrusted data in the form of code changes from Git commits. This creates a surface where instructions embedded within the code being reviewed could potentially influence the agent's behavior during the review process. While the skill is restricted to providing feedback and lacks capabilities for data exfiltration or file modification, this remains a characteristic of processing external content.
Ingestion points: The process involves reading file contents and commit history through commands like git diff-tree and reading related files for context.
Boundary markers: The instructions do not define specific delimiters or protective boundary markers for the code content being processed.
Capability inventory: The skill is scoped to read operations and local Git commands. It does not include network access, file-writing permissions, or administrative privileges.
Sanitization: There are no explicit instructions for sanitizing or filtering the content of the files before they are evaluated by the AI.

code-review