base44-platform-integrations-sdk

Pass

Audited by Gen Agent Trust Hub on Nov 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill exposes the agent to untrusted external data, creating a potential surface for indirect prompt injection.
  • Ingestion points: Untrusted data enters via InvokeLLM (through file_urls and add_context_from_internet), TranscribeAudio (audio_url), and ExtractDataFromUploadedFile (file_url) as described in references/mobile.md, references/web-spa.md, and references/web-tanstack.md.
  • Boundary markers: The documentation does not prescribe the use of specific delimiters to isolate external content within the agent's prompt context.
  • Capability inventory: The skill possesses powerful capabilities such as sending emails (SendEmail) and push notifications (SendPushNotification), which could be misused if the agent is compromised by malicious data.
  • Sanitization: The SDK documentation explicitly warns against routing untrusted input to email recipients and mentions that the backend performs content screening for abuse prevention.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 3, 2026, 02:21 PM