centrifugo

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill includes examples that place API keys and HMAC secrets directly into code/CLI snippets (e.g., curl X-API-Key header and jwt.encode with "<HMAC_SECRET>"), which encourages embedding real secret values verbatim in generated outputs and therefore creates an exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). In the Centrifugo real-time event proxy workflow, the server forwards client-submitted free-form fields (e.g., data, meta, RPC method/data, and optionally forwarded headers/body content) from unauthenticated/outsider clients to the configured backend endpoints (client.proxy.* and channel.namespaces[*].proxy.*) where an agent would ingest/interpret that text at runtime.