The Agent Skills Directory

[SAFE]: The skill serves as a configuration guide and template for RouterBase, focusing on standard AI gateway integration patterns.
[CREDENTIALS_UNSAFE]: No hardcoded credentials were found. The skill explicitly instructs against hardcoding keys and provides examples using environment variables (process.env.ROUTERBASE_API_KEY), which is a recommended security practice.
[PROMPT_INJECTION]: The skill includes a 'summarizeReleaseNote' function that interpolates untrusted user input directly into an LLM prompt. While this is a common design for the skill's purpose, it represents a surface for indirect prompt injection.
Ingestion points: text argument in summarizeReleaseNote function in SKILL.md.
Boundary markers: The template uses the standard OpenAI message structure (role: "user").
Capability inventory: External API calls via the RouterBase gateway.
Sanitization: No explicit sanitization or filtering of the input text is demonstrated in the snippet.
[REMOTE_CODE_EXECUTION]: The provided code examples use the standard openai library and do not involve downloading or executing untrusted remote scripts.

routerbase-model-gateway