The Agent Skills Directory

[SAFE]: The skill is comprised solely of markdown instructions that guide the agent's reasoning and communication style during code reviews.
[NO_CODE]: There are no scripts, environment configurations, or binary files included that would execute on a host system.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted code for review. 1. Ingestion points: User-provided code via the conversation context. 2. Boundary markers: Not defined in the skill text. 3. Capability inventory: No tools, scripts, or subprocess capabilities are requested or defined. 4. Sanitization: Not addressed. Because the skill defines no executable capabilities, the attack surface for indirect injection is minimal and confined to textual output.

code-review