talkdesk

Pass

Audited by Gen Agent Trust Hub on Apr 21, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Installs the @membranehq/cli package from the npm registry to interact with the Membrane platform.
  • [COMMAND_EXECUTION]: Executes commands via the membrane command-line tool to authenticate, manage connections, and perform Talkdesk operations.
  • [DATA_EXFILTRATION]: Accesses and processes Talkdesk data through a managed proxy service that handles authentication and credential management.
  • [PROMPT_INJECTION]: Potential for indirect prompt injection when processing external data from Talkdesk.
  • Ingestion points: Results from membrane action run and membrane request commands.
  • Boundary markers: None identified in the skill instructions.
  • Capability inventory: CLI command execution and network API requests.
  • Sanitization: No explicit sanitization or filtering of external content described in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 21, 2026, 04:11 PM