bluebubbles

Pass

Audited by Gen Agent Trust Hub on Mar 8, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect prompt injection surface detected. The skill describes an architecture for ingesting and processing untrusted messaging data from an external service.
  • Ingestion points: extensions/bluebubbles/src/monitor.ts acts as a webhook handler receiving external JSON payloads via an HTTP server.
  • Boundary markers: The instructions lack requirements for using delimiters or explicit 'ignore embedded instructions' warnings for the content of messages passed into the core reply pipeline.
  • Capability inventory: The skill's implementation includes network messaging capabilities (sendMessageBlueBubbles, buildBlueBubblesApiUrl), chat interaction tools, and file handling (downloadBlueBubblesAttachment).
  • Sanitization: While the skill mentions defensive normalization of sender and chat IDs, it does not include instructions to sanitize or escape message bodies before they influence agent behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 8, 2026, 12:25 AM