Agent Gateway
Your Copilot models. Your favorite coding agents.
Bring the models available in GitHub Copilot to Claude Code, Codex, Gemini CLI and more—through one gateway inside VS Code.
Website · Install for VS Code · Get started · Supported agents · Reference
Keep your agent. Choose your model.
Use the workflow you like without being tied to its default model provider.
Agent Gateway connects your coding agents to the models available in your
Copilot account. It handles API translation, agent configuration, model routing
and optional web search.
Claude Code · Codex · Gemini CLI · Antigravity CLI · Oh My Pi · Pi · Hermes · OpenClaw · Kilo Code · Cline · DeepSeek Harness · Cursor
Setup varies by agent; see supported agents for automatic
configuration, manual steps and live-testing limitations.
One gateway, more possibilities
| What you get |
|---|
| 🔌 Keep your tools | Your agent's shell commands, file edits and MCP tools stay with your agent. |
| ⚡ Set up from VS Code | Choose agents and a model, review the settings changes, then confirm. Backups and undo are built in. |
| 🔀 Speak the right API | Anthropic Messages, OpenAI Responses and Chat Completions, and Gemini APIs—including streaming and tool-call translation. |
| 🌍 Connect your machines | Share through a dev tunnel or add remote gateways. Choose a local or remote model as an explicit fallback. |
| 🔎 Search and read the web | Enable web search and URL fetch through Exa, Firecrawl, Tavily, Jina AI or You.com. Several engines offer anonymous access. |
| 🔍 See what's happening | Inspect models, remotes and recent errors in Status. Follow request IDs without logging prompts or keys. |
Runs inside VS Code. No separate gateway server to deploy. You need
GitHub Copilot Chat, signed in, and permission to access its language models.
Available models and usage limits depend on your account. The gateway runs
while VS Code is open; it does not bypass provider quotas.
Getting started
- Install Agent Gateway.
You need VS Code with GitHub Copilot Chat, signed in.
- When VS Code asks whether Agent Gateway may use the Copilot language
models, choose Allow. Without this, no models are served.
- Click Gateway in the status bar (or run Agent Gateway: Status) and
check that it says Ready.
- Run Agent Gateway: Configure Coding Agents, pick your agents and a
default model, review the changes and confirm.
Follow any remaining setup or restart instructions shown for your agents.
The default local endpoint is http://127.0.0.1:24444. Open Status >
API URLs to copy endpoints using your configured port, including token-count URLs.
Need a different setup? Connect an agent manually,
share models between machines, or
enable web search.
To turn the gateway off without uninstalling, set agentGateway.enabled to false.
Find the commands
Open the Command Palette (⇧⌘P on macOS, Ctrl+Shift+P on Windows/Linux)
and type Agent Gateway.
Explore
Supported agents · Model routing · Commands · Remote models · Web search · Settings · Troubleshooting · Privacy
Supported agents
| Agent | How it is configured |
|---|
| Claude Code | Configures the gateway and automatic model discovery. |
| Codex | Adds an agent-gateway provider with automatic model discovery and makes it the default. On macOS, enables Codex's native system proxy resolver to honor proxy exceptions. |
| Gemini CLI | Configures the endpoint, key, model and authentication. Restart the CLI; shell or project overrides may take precedence. Not yet verified live. |
Antigravity CLI (agy) | Configures Gemini mode and provides remaining setup steps. Requires a compatible model or mapping. |
| Oh My Pi (omp) | Adds the gateway as a provider with automatic model discovery; pick a model with /model. |
| Pi | Adds the gateway with its model list; run Configure Coding Agents again after models change. |
| Hermes Agent | Adds an agent-gateway provider; run hermes --provider agent-gateway or pick it in hermes model. |
| OpenClaw | Adds the gateway with its model list. Checked against OpenClaw's documentation, not yet tested with a live install. |
| Kilo Code | Adds the gateway with its model list; pick agent-gateway/<model> with /models. |
| Cline (CLI) | Adds an OpenAI Compatible provider and makes it active; switch models with cline -m <model>. The Cline VS Code extension is set up by hand. |
DeepSeek Harness (dsh) | Adds the gateway with its model list and makes your chosen model the default; removing it restores the previous default. |
| Cursor | Shows manual steps. Cursor calls models from its own servers, so it needs a shared (dev tunnel) URL. |
| Any other agent | Use the OpenAI, Anthropic or Gemini base URL; see manual setup. |
Each changed file is backed up next to itself as
<file>.agent-gateway-<timestamp>.bak. Configure Coding Agents >
Remove gateway reverts only what the gateway added.
Commands
All commands are under Agent Gateway in the Command Palette.
| Command | What it does |
|---|
| Status | Shows state, models, remotes, errors and copyable API URLs, with quick actions. |
| Restart Gateway | Restarts the local gateway. |
| Change Port | Moves the gateway to another port and offers to update your agents. |
| Set Fallback Model | Manages model mappings, AI suggestions and an optional local or remote fallback. |
| Configure Coding Agents | Sets up, updates or removes the gateway in your agents. |
| Web Search and Fetch | Chooses the web search engine and its API key. |
| Share Gateway via Dev Tunnel | Shares your gateway with another machine and manages the gateway API key. |
| Manage Remote Upstreams | Adds, tests or removes another machine's gateway. |
Share models between machines
On the machine with Copilot, run Share Gateway via Dev Tunnel. It installs
and signs in to the devtunnel CLI if needed, offers to create a gateway API
key, and starts a tunnel whose URL stays the same across runs. Hosting runs in
an extension-managed background process, not an integrated terminal. Use Copy
client setup to send the URL (the key is never included).
The gateway checks the public connection automatically. Status shows
Connecting, then Connected with the discovered model count, or the
connection-check failure. You can start, manage or stop sharing from Status.
Run Share Gateway via Dev Tunnel again and choose Stop sharing to stop
hosting. Hosting also stops when the extension shuts down. Installation and
sign-in may still open a terminal for interactive steps.
On the other machine, run Manage Remote Upstreams > Add remote
upstream, paste the URL and enter the key. Its models appear as
<name>--<model>.
Anyone with the tunnel URL and key can use your Copilot models. Always set a
gateway API key before sharing.
Web search and URL fetch
Web search is off until you turn it on. Run Web Search and Fetch, pick an
engine, and enter its API key, or leave the key empty to use the engine's
free anonymous endpoint where there is one.
| Engine | Without a key |
|---|
| Exa | Search and fetch (rate-limited) |
| Firecrawl | Search and fetch (rate-limited) |
| Jina AI | Fetch only |
| Tavily | Needs a key |
| You.com | Needs a key |
| Brave Search | Needs a key (search only) |
| Kagi | Needs a key |
Keys are stored in VS Code secret storage, never in settings.json. For Codex,
the command also offers to turn on its live search (web_search = "live").
Search queries, results and page text are never logged.
Settings
| Setting | Default | Description |
|---|
agentGateway.enabled | true | Run the gateway. |
agentGateway.port | 24444 | Local port. Use Change Port to change it safely. |
agentGateway.logLevel | info | error, warn, info or debug. |
agentGateway.remoteUpstreams | [] | Other machines' gateways (managed by Manage Remote Upstreams). |
agentGateway.webSearch.provider | none | Web search engine (managed by Web Search and Fetch). |
agentGateway.webSearch.maxSearchesPerRequest | 5 | Most searches per request (1-20). |
agentGateway.webSearch.model | "" | Model for Gemini CLI search requests when its model is unknown. |
agentGateway.modelMappings | Two empty mappings¹ | Map requested names to local models through Set Fallback Model. |
agentGateway.fallbackModel | "" | Model for unknown model names (managed by Set Fallback Model). |
¹ The default mapping keys are claude-haiku-5.5 and claude-sonnet-5, both
empty. They do not force a replacement model.
Model mappings and fallback
Model selection chooses the model your agent requests. Model mappings
route a requested name to another model through agentGateway.modelMappings.
Choose your main model in Configure Coding Agents. For sub-agents and
background tasks that request a different model name, use Set Fallback Model
to manage routing in the gateway—no manual agent configuration is needed for
these mappings.
- Map a specific name: choose a Mapping entry and select its target.
- Get suggestions: choose Initialize model mappings with AI…, review
the suggestions and save only the mappings you want.
- Handle unknown names: select a local or remote fallback model, or
No fallback to disable it. No fallback is selected automatically.
Existing-model errors do not trigger fallback. Gemini unknown names use model mappings,
not the general fallback rule. See model routing details.
AI-assisted model mappings
Choose Set Fallback Model > Initialize model mappings with AI…. Enter the
requested model names, choose an advisor model, then review and select suggestions
to save. Existing mappings stay unchanged unless you select replacements.
Canceling saves nothing; saved mappings apply without restarting the gateway.
The advisor receives model names and metadata—not files or conversations—and
uses Copilot quota. Suggestions target compatible local models, not remote models;
they are recommendations, not benchmark rankings.
See mapping details for defaults,
matching rules and configuration precedence.
Server tools
Your agent's own tools (commands, file edits, MCP servers) run on your
machine and work unchanged.
Some tools normally run on the model provider's servers. Copilot models in VS
Code have no such servers, so the gateway runs what it can itself:
| Server tool | Through the gateway |
|---|
Web search (Anthropic web_search, OpenAI web_search, Gemini googleSearch) | Run by the gateway through the engine you pick in Web Search and Fetch. Off until you pick one. |
Gemini URL context (urlContext, Gemini CLI's web_fetch) | Run by the gateway, unless the same request also has the agent's own tools. |
Anthropic web_fetch | Run by the gateway with the same engine. Only pages whose URL is already in the conversation are fetched. |
Code execution (Anthropic code_execution, OpenAI code interpreter, Gemini codeExecution) | Not supported. Left out (Anthropic: the request fails). |
OpenAI file_search, image generation, remote mcp, hosted tool_search; Gemini Maps and File Search | Not supported. Left out. |
| Other Anthropic server tools (MCP connector, tool search, advisor) | Not supported. The request fails. |
A tool that is left out is never shown to the model, and the Agent Gateway
output panel notes it. Support depends on the tool requested, not just the agent.
Troubleshooting
- Status shows Waiting for Copilot: make sure Copilot Chat is signed in and
that you allowed Agent Gateway to use the language models.
- Port in use: another program has port 24444. The gateway keeps retrying;
use Change Port to pick another one.
- A request failed: every response has a
request-id (and
x-request-id) header. Find that
ID in Status > Open error log or in the Agent Gateway output
panel. Set agentGateway.logLevel to debug for more detail.
- An agent stopped working after changing the port: run Configure Coding
Agents again.
Known issues
- Codex browser tools need separate authentication: model requests can work
through the gateway while browser tools require a Codex sign-in. The gateway
does not provide that token. See Codex authentication.
- Antigravity CLI fails with "invalid stream chunk: : ping": during a long
pause the gateway sends a standard keep-alive line on the stream, and
Antigravity CLI 1.4.3 does not accept it. Other agents are not affected.
Retry the request.
- Claude Code auto-mode eligibility notice: run Configure Coding Agents
again to apply the gateway-compatible setting. Checks may use Copilot quota
rather than the no-charge classifier path.
Privacy and security
- The gateway listens on your computer only (loopback). Nothing is reachable
from other machines unless you share it through a dev tunnel.
- Prompts, request bodies, tool arguments, search queries and API keys are
never logged.
- Gateway and search-service keys are managed in VS Code secret storage.
Configuring an agent may also write the gateway key into its settings or
environment file. Treat these files and their backups as sensitive.
More information
- Reference: API endpoints, manual client setup, how web
search and model fallback work, logs.
- Contributing: building, testing and the project's design.
- Changelog.
Star history
License
MIT
