3scribe

Pass

Audited by Gen Agent Trust Hub on May 1, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructions include installing the @membranehq/cli package from the NPM registry. This is an expected dependency provided by the vendor to facilitate the integration.- [COMMAND_EXECUTION]: The skill utilizes the membrane command-line interface to perform operations such as authentication, action discovery, and running transcription jobs.- [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it retrieves and processes content from the 3Scribe service.
  • Ingestion points: Data returned from transcription jobs and job searches via fetch-transcript and search-transcription-jobs.
  • Boundary markers: There are no explicit delimiters or instructions provided to separate the retrieved transcription content from the agent's internal instructions.
  • Capability inventory: The skill has access to the membrane CLI which can perform network requests and execute arbitrary actions within the Membrane platform.
  • Sanitization: The skill does not define any sanitization or filtering logic for the data ingested from the external service.
Audit Metadata
Risk Level
SAFE
Analyzed
May 1, 2026, 10:23 PM
Security Audit — agent-trust-hub — 3scribe