amplitude-automation

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the way it handles external data.
  • Ingestion points: The skill retrieves potentially untrusted data from Amplitude through tools like AMPLITUDE_GET_USER_ACTIVITY, AMPLITUDE_FIND_USER, and AMPLITUDE_GET_COHORT as described in SKILL.md.
  • Boundary markers: There are no instructions to use delimiters or security markers to differentiate ingested data from system instructions.
  • Capability inventory: The agent has the ability to perform write operations, including AMPLITUDE_SEND_EVENTS, AMPLITUDE_IDENTIFY, and AMPLITUDE_UPDATE_COHORT_MEMBERSHIP.
  • Sanitization: The skill instructions do not specify any sanitization, validation, or escaping of the content retrieved from Amplitude before it is processed.
  • [EXTERNAL_DOWNLOADS]: The setup instructions require the addition of an external third-party MCP server located at https://rube.app/mcp, which is outside the list of trusted vendors.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 06:58 AM