openclaw-task-worker

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill prompt includes a hardcoded Supabase anonymous JWT in plain text, which is an insecure practice as it embeds a live secret credential directly into the prompt context.

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

• Third-party content exposure detected (medium risk: 0.30). The OpenClaw task worker skill polls a distributed task network where any third party can publish tasks containing untrusted free text payloads ingested by the agent.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs the agent to dynamically poll, register with, and execute tasks fetched from an untrusted third-party Supabase backend via curl commands, constituting an unknown/suspicious operational dependency.