EXEMPLARExemplar is part of the NVIDIA Inception program.
NVIDIA InceptionExemplar is part of the Google for Startups program.
Google for StartupsMarshal · LLM, MCP & agent routing
Gateway
A unified gateway for LLM providers, auto MCP tools from 50+ connectors, and agent runtimes—with one identity, policy layer, and audit trail
Marshal · Gateway
How Gateway works
A unified gateway for LLM providers, auto MCP tools from 50+ connectors, and agent runtimes—with one identity, policy layer, and audit trail
AI GATEWAY
/gateway/v1
Requests flow in with one org key (sky rails); the gateway routes to providers and streams telemetry out (orange rails).
Requests flow in with one org key (sky rails); the gateway routes to providers and streams telemetry out (orange rails).
Why it matters
Built for production agents—not prompt playgrounds
How this harness capability fits the Exemplar platform: one identity, one policy layer, and one audit trail across console, MCP, and SDK.
Why Exemplar
Fragmented integrations mean every agent team rebuilds bridges to AWS, Linear, Snyk, and internal APIs—with different auth and no shared audit.
Exemplar’s gateway is the standardized bridge: 50+ connectors become MCP tools automatically, whether the caller is a human in the console or an agent in any framework.
What Exemplar delivers
Out-of-the-box MCP and API integrations anchored to catalog and Day 2 Ops—tools generated from connectors, not hand-wired per client.
Framework-agnostic agent entry: LangGraph, CrewAI, AutoGen, ADK, and custom runtimes share the same governed gateway.
LLM routing with policy hooks—model choice, fallbacks, and spend caps at the gateway—not buried in client config.
How teams use it
Enable connectors; MCP tools appear for IDE clients and agent frameworks without per-integration glue code.
Register custom agents and sub-agents; orchestration routes tool calls through the gateway with structured allow/deny responses.
Capability checklist
Developer guide
Official documentation on docs.exemplar.dev for this capability.
Open developer guide (opens in a new tab)Get started
Put Gateway to work with the rest of Marshal—or talk to us about your agent stack.
From the blog
Related posts on exemplar.dev.
- Best AI Agent Control Plane & Harness Tools in 2026
The best AI agent control plane and harness tools for running agents in production — governing actions, managing token costs, orchestrating workflows, and keeping a full audit trail. Compared and ranked.
- Agents, context, and guardrails on a unified platform
From code completion to production actions; Context Lake, catalog, governance, and DevX Assist/MCP for safe automation.
- What is Agentic DevOps? The Complete Guide
Agentic DevOps uses AI agents to execute DevOps and operational tasks autonomously — provisioning, incident response, secret rotation, and more — within a governed framework. Definition, use cases, safety requirements, and how to get started.