openclaw-task-worker
Audited by Snyk on Apr 1, 2026
HIGH W007: Insecure credential handling detected in skill instructions.
- Insecure credential handling detected (high risk: 1.00). The prompt includes an actual ANON token and instructs embedding worker_api_key values verbatim into curl JSON bodies/headers (e.g., "worker_api_key": "sk_xxx"), which requires the LLM to handle and output secret values directly, creating an exfiltration risk.
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches publisher-provided, user-generated task content (full "content", summaries, images, platform_config) from the OpenClaw Supabase API (https://ieoybuwlfiwbownpwpqc.supabase.co) via get_task/claim_task and requires the agent to read and act on that untrusted content to perform publishing, which could enable indirect prompt injection.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 1.00). The skill performs runtime POST requests to https://ieoybuwlfiwbownpwpqc.supabase.co (BASE) to claim tasks and receive the task "content" (Markdown) that directly instructs the agent what to do, so this external URL supplies runtime instructions that control the agent.
HIGH W008: Secret detected in skill content (API keys, tokens, passwords).
- Secret detected (high risk: 1.00). The ANON value in the Quick Start block is a full, high-entropy JWT: eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJzdXBhYmFzZSIsInJlZiI6Imllb3lidXdsZml3Ym93bnB3cHFjIiwicm9sZSI6ImFub24iLCJpYXQiOjE3NzA5NzA3OTIsImV4cCI6MjA4NjU0Njc5Mn0.2Yfe-IMMF5s2lAILMS7zIFI1CK1X6nrb-IU8Tpo_9Kk
It is assigned to ANON and explicitly used as the apikey header in curl examples to call the Supabase REST endpoints, so it is a directly present, usable credential. This is not a placeholder, truncated, or a simple setup password — it is high-entropy and provides service access, so it should be flagged.
Ignored items: the BASE URL is just an endpoint (not a secret). Occurrence of "sk_xxx" in examples is a placeholder and intentionally ignored per the rules.
Issues (4)
Insecure credential handling detected in skill instructions.
Third-party content exposure detected (indirect prompt injection risk).
Unverifiable external dependency detected (runtime URL that controls agent).
Secret detected in skill content (API keys, tokens, passwords).