Skip to main content
Remote Agents are available on both SaaS and Self-hosted deployments.
Already have agents running elsewhere? Register them here. Your organization likely has agents running on Bedrock, Vertex AI, Azure AI, LangGraph, or custom infrastructure. Agent Registry lets you connect all of them into a single control plane — without moving or rewriting anything. Once a remote agent is registered, it gets the same treatment as any TrueFoundry-native agent:

Single Discovery Surface

All agents — regardless of where they run — are visible in one place. Teams stop asking “where’s that agent?”

RBAC

Control which teams and users can discover and invoke each agent.

Policies

Apply guardrails, content filters, and governance policies uniformly across all remote agents.

Unified Metrics

Get request counts, latency, error rates, and usage data across every agent from a single pane of glass.
Agent Registry showing registered Remote agents

​
Supported Frameworks

Remote agents are registered with a framework type that determines how TrueFoundry communicates with the agent. Two framework types are supported:
FrameworkProtocolDescription
A2AAgent-to-Agent (A2A)Google’s open interoperability standard. Agents communicate via JSONRPC, expose an agent card at a well-known path, and support structured message exchange. Recommended for new agents.
CustomHTTPProxy any HTTP-compatible agent endpoint. No protocol constraints — TrueFoundry forwards requests to your endpoint as-is.

​
Registering a Remote Agent

On the Agent Registry page, click Create New Agent and select Integrate a Remote Agent.
Create New Agent button on the Agent Registry page
Create New Agent dialog with Integrate a Remote Agent highlighted
Fill in the registration form:
1

Basic information

Provide the agent’s Name, Description, and optional Tags.
2

Collaborators

Add team members or groups with access. Each collaborator gets either Agent Manager (can edit the agent) or Agent Access (can invoke the agent) permissions.
3

Endpoint URL

Enter the URL where the agent is reachable. TrueFoundry will proxy all requests to this endpoint.
4

Framework

Select A2A or Custom. For A2A agents, specify the Agent Card Path (defaults to /.well-known/agent-card.json).
5

Authentication

Optionally configure auth for requests to the remote agent:
  • Header Auth — Static API key or token sent via request headers (e.g., Authorization: Bearer <token>).
  • Token Passthrough — Forward the caller’s token to the remote agent as-is.
Remote agent registration form showing name, description, collaborators, and endpoint URL
Remote agent registration form showing framework, agent card path, and authentication options

​
Registration Fields Reference

FieldRequiredDescription
NameYesUnique identifier for the agent within your organization.
DescriptionYesShort description of the agent’s purpose.
URLYesThe endpoint URL where the agent is reachable.
FrameworkYesA2A or Custom.
Agent Card PathA2A onlyPath to the agent card JSON (default: /.well-known/agent-card.json).
CollaboratorsYesUsers or teams with agent-access or agent-manager roles.
Auth DataNoHeader Auth (static headers) or Token Passthrough.

​
A2A Agents

The Agent-to-Agent (A2A) protocol is an open standard for agent interoperability — a common language that lets agents from different frameworks and vendors communicate with each other. Agent Registry natively supports registering A2A-compatible agents. If your agent speaks A2A, register it and it immediately becomes part of your organization’s agent ecosystem — discoverable, governable, and composable with other agents. This matters as the agent ecosystem matures. Agents built by different teams, on different platforms, using different frameworks will need to interoperate. A2A is how that happens. And Agent Registry is where all of them come together.

​
Invoking an A2A Agent

Once registered, the agent appears on the Agent Registry page. Click the code snippet button to get a ready-to-use snippet for invoking the agent via the A2A Python SDK.
Agent Registry with code snippet button highlighted on a Remote agent
The generated snippet handles agent card resolution, client setup, and message exchange — all routed through TrueFoundry’s proxy with your configured authentication.
Python code snippet using A2A SDK to invoke a registered remote agent
If you’re building new agents — internal or external — consider building them A2A-compatible. It makes them registerable, composable, and interoperable by default.

​
Monitoring

​
Traces

Every request to a remote agent is traced through TrueFoundry. Navigate to AI Monitoring > Request Traces to inspect individual requests, including the full JSONRPC request/response payload for A2A agents.
Request Traces for a remote A2A agent showing JSONRPC request and response

​
Metrics

Agent-level metrics — request counts, latency percentiles, error rates, and failure breakdowns — are available under AI Monitoring > Metrics > Agent Metrics. Filter by agent name to drill into a specific remote agent’s performance.
Agent Metrics dashboard showing request volume, latency, and failure rates