The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides a code template for a Next.js Route Handler that interpolates untrusted user input directly into a summary prompt. This creates a surface where a malicious user could provide input designed to override the agent's instructions.
Ingestion points: app/api/ai/route.ts (retrieves text from req.json())
Boundary markers: Absent; input is appended after a newline.
Capability inventory: Performs a network request via the OpenAI SDK to a remote gateway.
Sanitization: None; raw input is used in the template literal.
[EXTERNAL_DOWNLOADS]: The skill instructs users to install the openai package from the official NPM registry, which is a standard industry library.
[COMMAND_EXECUTION]: The instructions include a curl command template for users to manually query available models from the vendor's API gateway (factchat-cloud.mindlogic.ai). This is a legitimate documentation pattern for API interaction.

factchat-gateway