Engineering Agents
Let developers choose. Enforce policy everywhere.
Your engineers use Claude Code, Cursor, Copilot, Codex, or any MCP-compatible agent. Guard applies the same runtime policy across all of them — without a separate configuration per tool.
Developers keep their tools.
Guard integrates through the CLI hook, the HTTP proxy, and the MCP layer. Developers install once and keep using whichever AI tool they prefer. Policy travels with the agent, not with the tool.
AGENT TOOLS
- Claude Code
- Cursor
- Copilot
- Codex
CLIENTS
- Claude DesktopMCP
- ChatGPTMCP
- CursorMCP
- Custom agentsproxy / MCP
BYO GATEWAYS
- Azure OpenAI
- OpenRouter
- Portkey
- Helicone
- LiteLLMPreview
- ConductAI native
Consistent policy across every agent tool.
The same policy file governs all four agent tools. A rule written for Claude Code applies to Cursor and Codex without modification. One policy, no per-tool exceptions.
Same rule — all surfaces
Allow, approve, or block — per action.
Guard applies the same decision framework across all agent tools. Routine actions proceed without friction. Consequential ones pause for approval. Prohibited ones stop.
What this changes for engineering leads
Before Guard
- Each AI tool enforces policy differently — or not at all
- Audit trails scattered across tools
- Policy drift as teams adopt new agents
- Compliance questions answered by guessing
With Guard
- One policy file governs all agent tools
- Every decision logged in one audit trail
- New agents inherit policy automatically
- Compliance questions answered with signed receipts