The Agent Skills Directory

[SAFE]: The implementation handles sensitive OWS API keys using environment variables (OWS_TENANT_<SLUG>_API_KEY) and resolves them server-side, ensuring credentials never reach the frontend.
[SAFE]: Incoming webhooks in references/src/webhooks.ts are protected by mandatory HMAC-SHA256 signature verification using timingSafeEqual to mitigate timing attacks.
[SAFE]: The OWS proxy in references/src/proxy.ts enforces an allow-list (ISSUER_PERMITTED_PREFIXES) that restricts requests to specific credential issuance endpoints, adhering to the principle of least privilege.
[SAFE]: The skill uses well-known Node.js dependencies and communicates exclusively with trusted vendor domains (*.igrant.io) for its API operations.
[INDIRECT_PROMPT_INJECTION]: The skill exposes a data ingestion surface via the /webhook endpoint which accepts and stores external JSON data. While protected by HMAC verification, the stored data is served via SSE without explicit sanitization, which represents a potential injection surface if the data is processed by downstream LLM-based systems.
Ingestion points: references/src/webhooks.ts (POST /)
Boundary markers: HMAC signature verification is implemented in verifySignature using a shared secret.
Capability inventory: In-memory event storage and SSE streaming (references/src/sse.ts); capability to proxy requests to OWS.
Sanitization: None; data is stored and emitted as received after HMAC validation.

igrantio-issuer-backend