The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted source code as part of its core code review functionality. This creates a potential surface for indirect prompt injection where instructions embedded in code comments or strings could attempt to influence the agent's feedback.
Ingestion points: The skill processes external code changes, patterns, and architectural decisions as specified in the development workflow within SKILL.md.
Boundary markers: The instructions do not specify explicit delimiters, XML tags, or specialized guardrails for the untrusted code input to prevent it from being interpreted as instructions.
Capability inventory: The skill has no tools defined in its allowed-tools configuration, meaning it lacks the ability to execute commands, write files, or make network requests, which significantly limits the impact of any potential injection.
Sanitization: There are no instructions for sanitizing, escaping, or filtering the content of the code or data being reviewed.

code-reviewer