twilio-conversation-memory
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill facilitates the storage and retrieval of conversation context, creating a surface for indirect prompt injection. Malicious instructions embedded in conversation transcripts or user-submitted observations could be stored in the Memory Store and later retrieved during a 'Recall' operation, potentially influencing the agent's logic in subsequent turns.
- Ingestion points: Data retrieved via the
RecallandObservationsAPI endpoints described inSKILL.md. - Boundary markers: None are specified to help the agent distinguish between recalled memory and current instructions.
- Capability inventory: The skill performs network operations via the Twilio API and manages persistent conversation data.
- Sanitization: No specific sanitization or filtering logic is provided for the content of retrieved observations or summaries.
- [SAFE]: The skill uses official vendor domains (memory.twilio.com) and recommends standard industry practices for credential management, such as using environment variables (
TWILIO_ACCOUNT_SID,TWILIO_AUTH_TOKEN) rather than hardcoding secrets.
Audit Metadata