The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines an interface for handling untrusted external data via HTTP requests.
Ingestion points: The OpenAPIHandler.handle method in SKILL.md processes incoming HTTP requests including query parameters, path segments, and request bodies.
Boundary markers: The instructions consistently demonstrate the use of Zod schemas to define strict boundaries for all inputs.
Capability inventory: The skill defines procedure handlers; provided examples are limited to data retrieval and simple response generation.
Sanitization: The skill recommends utilizing the SmartCoercionHandlerPlugin and Zod schemas to ensure all external data is validated and coerced into safe, expected types.
[EXTERNAL_DOWNLOADS]: The skill references official documentation and resource files from orpc.dev. These references are informative and point to the legitimate project website.
[SAFE]: Comprehensive analysis of the instructions and code snippets revealed no evidence of prompt injection, data exfiltration, or unauthorized system access.

orpc-openapi