cloud-onboarding

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill translates natural language requests into administrative actions on Elastic Cloud, creating an interface for managing identities and access.
  • Ingestion points: Processes user requests for access management in SKILL.md (e.g., "add Alice to my search project").
  • Boundary markers: Implements a structured decomposition process and mandates existing state verification to ensure requested actions are appropriate.
  • Capability inventory: Utilizes the elastic CLI to perform organization member management, role assignments, and API key management.
  • Sanitization: Explicitly requires human confirmation for destructive actions like revoking members or keys and prohibits secret sharing in the chat interface.
  • [COMMAND_EXECUTION]: Orchestrates actions through the official elastic CLI tool, which is a trusted vendor resource.
  • Evidence: SKILL.md maps HTTP shorthand operations to CLI commands such as elastic cloud orgs and elastic es security.
  • [EXTERNAL_DOWNLOADS]: Interacts with official vendor API endpoints for control-plane operations.
  • Evidence: Communicates with https://api.elastic-cloud.com as documented in references/api-reference.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 05:09 PM
Security Audit — agent-trust-hub — cloud-onboarding