ag2-acp
Pass
Audited by Gen Agent Trust Hub on Jul 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The documentation instructs users to install external CLI adapters from the official npm registry (e.g., @agentclientprotocol/claude-agent-acp) and the ag2 library via pip.- [COMMAND_EXECUTION]: The skill's primary function is to spawn and manage subprocesses for external CLI agents. These processes are granted access to the local terminal and file system as part of their intended coding tasks.- [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection because the driven coding agents process arbitrary codebase content. Malicious code within a project could attempt to influence the agent's behavior.
- Ingestion points: Files within the specified workspace directory (cwd).
- Boundary markers: Not explicitly defined for external file contents.
- Capability inventory: Access to file system operations (fs/*) and terminal commands.
- Sanitization: Implements a permission_policy system (ask/auto/deny) that allows for human-in-the-loop (HITL) approval or automated gating of sensitive agent actions.
Audit Metadata