managed-agent

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill facilitates the execution of shell commands both locally through the Bash tool and within the remote Anthropic cloud environment as part of its core functionality.
  • [EXTERNAL_DOWNLOADS]: Provisions remote environments with software packages from public registries and connects to external MCP servers via URL, which is consistent with the management of a well-known technology service.
  • [PROMPT_INJECTION]: As the skill processes user-supplied messages and transcript data from remote agents, it possesses an indirect prompt injection surface.
  • Ingestion points: The message parameter in managed_agent_prompt and the raw transcript output from managed_agent_events (SKILL.md).
  • Boundary markers: No explicit instructions for delimiters or escaping are provided in the skill instructions.
  • Capability inventory: Access to local Bash and control over remote cloud agent lifecycles via the managed_agent_* toolset.
  • Sanitization: No specific sanitization or validation of external content from the message or event logs is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 01:22 PM
Security Audit — agent-trust-hub — managed-agent