agenthub

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt instructs the agent to ask for and include API keys (e.g., Authorization: Bearer $AGENTHUB_API_KEY and the validation_api_key field) directly in API requests/payloads, which requires the LLM to handle and output secret values verbatim and creates exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.75). Workflow 3b (“Create a task from PDF/DOCX/PPTX/Audio URL”) calls POST /api/ingest/extract on a user-supplied external URL, and the extracted document text is then sent as program_md into POST /api/tasks, which becomes LLM-readable context for later patch generation.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to GET the enriched prompt from the platform (e.g. https://agenthub.clawsy.app/api/tasks/{id}?enriched=true) and "Use the enriched_prompt as your system instructions", so remote content from https://agenthub.clawsy.app is fetched at runtime and directly controls the agent's prompts.