The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The file references/patterns.md contains a hardcoded string "sk-abc123secretkey". This is explicitly used within an educational example of a "BAD" coding anti-pattern to demonstrate how secrets should NOT be stored in source code.
[INDIRECT_PROMPT_INJECTION]: This skill operates by analyzing external repository content, which creates a potential surface for indirect prompt injection if the code being reviewed contains adversarial instructions. 1. Ingestion points: Files within the repository being reviewed. 2. Boundary markers: None defined in the skill instructions to distinguish between analysis instructions and data content. 3. Capability inventory: The agent is authorized to inspect repository context, modify files, and execute shell commands (build, test, lint, validation). 4. Sanitization: There are no instructions for sanitizing or escaping the content of the reviewed code before processing.

code-review