The safeguards map / 3.3 Gateways
agentgateway
Drafted with AI
Drafted with AI from the cited sources under the direction of UnGovr staff. UnGovr holds editorial responsibility for this page. How this site is made
An open-source proxy for service, model, Model Context Protocol (MCP) and agent-to-agent traffic, with per-call authorization and tracing.
Where it sits
3.3 Gateways (primary): Sits on the request path between agents and their models, MCP tools and other agents, and allows or denies each call by its own CEL rules or by asking an external authorization server.
Route, secure, observe, and govern services, LLM provider traffic, MCP tools, and agent-to-agent communication without stitching together separate gateways.
captured
When authorization decisions need to be made out-of-process, use an external authorization policy.
captured
Route-level HTTP policies, such as `authorization` and `localRateLimit`, can also read `mcp.sessionId` and `mcp.tool.arguments` when the request arrives.
captured
Facts
- Concern
- security
- Names the law
- no
- Subject
- agentAI system
- Used by
- those who run it
- Runs
- at runtime
- Policy comes from
- owner-entered settingsan external decision point
- Traffic it sees
- model callstool calls (MCP)agent to agent (A2A)
- Can the agent bypass it
- depends
Handoffs
produces Runtime records: OpenTelemetry traces (OTLP, GenAI semantic conventions)
Every quote is verbatim from a stored capture of the page linked under it. Placement is our reading of the product's own documentation; no publisher has reviewed or endorsed it. Rows CC BY 4.0. Names, marks and logos belong to their owners; logos identify each product and are not part of the dataset.