direct-messages
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to interact with and process X/Twitter Direct Messages, which are user-controlled inputs. This creates a vulnerability surface where an attacker could send DMs containing instructions designed to hijack the AI agent's behavior. \n
- Ingestion points: X/Twitter Direct Message inbox and message requests on
x.com/messages. \n - Boundary markers: The documentation does not specify the use of delimiters or instructions to ignore embedded content within the messages. \n
- Capability inventory: The skill can send personalized messages, export message data, and manage conversation flows. \n
- Sanitization: No sanitization or filtering measures are described to handle untrusted text content within the processed messages. \n- [NO_CODE]: The skill references external JavaScript implementation files (
src/sendDirectMessage.jsandsrc/dmManager.js) that were not included in the provided analysis set. The security posture of the actual logic used to send and manage DMs cannot be verified, which is significant given the instructions to manually execute these scripts in a sensitive browser environment.
Audit Metadata