nanoclaw-traffic-guardian

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

• Third-party content exposure detected (high risk: 0.70). The skill's required behavior explicitly says it will inspect HTTP request/response text (see SKILL.md "Required First Implementation Behavior" item 4 and SPEC.md "Inspect HTTP request/response text up to a bounded byte limit"), which means the runtime will ingest untrusted third‑party network content (external HTTP/HTTPS traffic) that the agent reads and uses to produce findings via MCP tools, creating a clear channel for indirect prompt‑injection from user-generated or external web content.