power-automate-governance

Pass

Audited by Gen Agent Trust Hub on Apr 26, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes flow descriptions provided by makers.
  • Ingestion points: Flow descriptions and tags retrieved via list_store_flows and get_store_flow in SKILL.md.
  • Boundary markers: No delimiters are used to isolate untrusted user data in the analysis workflows.
  • Capability inventory: The skill can stop flows using set_store_flow_state and modify governance data via update_store_flow.
  • Sanitization: No input validation is performed on the data fetched from Power Automate.
  • [COMMAND_EXECUTION]: The skill includes capabilities to administratively stop Power Automate flows.
  • Evidence: The set_store_flow_state tool allows the agent to disable flows as part of archive or offboarding workflows.
  • [EXTERNAL_DOWNLOADS]: The skill connects to the external service FlowStudio for its operations.
  • Evidence: References to https://mcp.flowstudio.app for API access and documentation in SKILL.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 26, 2026, 10:29 AM
Security Audit — agent-trust-hub — power-automate-governance