What does ProxyPilot do?
ProxyPilot runs a local Mac control layer so Xcode Agent requests can route to the upstream endpoint you choose, including cloud APIs, local Ollama or LM Studio servers, and custom OpenAI-compatible endpoints.
Endpoint routing for Xcode agents
Route Xcode Agent work to the model you actually want.
v1.16.2
CLI v1.16.2
16 providers
ProxyPilot is a Mac app, headless CLI, and MCP control layer for Xcode Agent traffic. v1.16.2 adds Moonshot.ai (Kimi) as a built-in provider, rebuilds setup around Xcode, and keeps your installed CLI and RepoGPS current when the app updates.
$ curl -fsSL https://micah.chat/downloads/proxypilot-install.sh | bash
What's new in v1.16.2
Moonshot.ai, an Xcode-first setup path, and companions that update with the app
Moonshot.ai (Kimi) joins the built-in providers, with Kimi K3 on Moonshot's native Messages API. Setup now leads with Xcode, and updating the app also updates your installed ProxyPilot CLI and managed RepoGPS.
v1.16.2
Moonshot.ai, an Xcode-first setup path, and companions that update with the app
Moonshot.ai (Kimi) is now a built-in provider with its own key setup, model discovery, and CLI catalog entry. Kimi K3 routes through Moonshot's native Messages API, so thinking signatures and tool-use content survive the round trip, while other Moonshot models continue to use the OpenAI-compatible translator. Setup is rebuilt around Xcode: the sidebar leads with Xcode Setup, then Connections, then optional RepoGPS, with three direct setup shortcuts and a reusable help guide, and the Terminal tour no longer opens on its own. Updating the app now also updates an already-installed ProxyPilot CLI and managed RepoGPS through staged, hash-verified replacement that preserves newer versions and package-manager links; protected copies can be updated with administrator authorization from Advanced. Session status separates background proxy activity from the client actually using it, ended RepoGPS sessions release Home and the toolbar, and Apply Models & Restart stops an existing background proxy through the verified path and recovers the listener after a port conflict.
v1.15.1
Safer local routing, stronger privacy, and clearer session usage
Cloud-backed inference now requires a separate, automatically managed local credential, while browser-origin requests are rejected before reaching a provider. Managed Xcode settings migrate automatically; credential-free local models remain passwordless. Encrypted logs marked for deletion on quit are removed before the app exits, and Exacto model selections survive synchronization. RepoGPS gains selected-versus-applied model details, known model limits and pricing, durable per-model and per-role usage, and bounded independent requests that leave the main route unchanged. The experimental GitHub Copilot integration has been removed; independently installed helpers are left untouched.
v1.14.0
The RepoGPS coding harness, a live cockpit, and optional analytics
ProxyPilot now bundles RepoGPS, a coding harness you install and launch from inside the app; managed installs are verified and applied atomically, then receive signed RepoGPS updates without reinstalling ProxyPilot. You can also check or apply those updates directly with rgps update --check and rgps update. Home became a live cockpit that recognizes an in-flight harness session and reports the route actually serving it, rather than the one merely configured. A guided four-step tour installs the harness and opens a project as you go, a built-in glossary explains the vocabulary it uses, and a marker beside the version number points to what changed. Optional analytics are default off with prospective consent and no backfill of earlier activity, and session usage is now attributed to the harness session that made the requests, so an empty report can no longer be mistaken for an idle session.
v1.13.3
Authenticated updates, tighter routing boundaries, and resilient local data
CLI self-updates now verify both the exact artifact checksum and its Ed25519 signature before installation. Local proxy authentication and model allowlists fail closed when stored credentials are in play, Agent runtime packages are verified before installation, and streaming, Session History, logs, exports, sidecars, and secrets now have stronger bounds, sanitization, permissions, and cleanup.
v1.13.2
Direct Agent setup, accurate route labels, and stronger streaming
The Home dashboard's manual-registration action now opens the Proxy page at a visible, highlighted setup panel containing the exact Xcode Agent values. Home and menu-bar labels distinguish the configured route from the route actually serving requests. Stream framing is more standards-compatible across providers, unavailable routing controls stay hidden until their local companion is installed, and Advanced preferences no longer show an unused update-channel option.
v1.13.0
Live route switching, cache-aware estimates, and I/O logging fixes
A new proxypilot route command family switches the active provider live, verified against stored credentials and rolled back automatically if the new route fails, while the reserved proxypilot-active model alias gives agents one stable model ID across switches. New capabilities, providers, runtime, and telemetry commands expose machine-readable JSON contracts, and MCP gains a proxy_route_set tool. Cost estimates use OpenRouter's cached-read and cache-write pricing, Input & Output logging settings apply live to CLI-started proxies, and ProxyPilot Agent mode shows the same model picker and routing status as Claude Agent mode.
v1.12.1
Icon-derived brand mark, Layout menu, and I/O logging fix
An icon-derived brand mark now sits in the toolbar next to the traffic lights, with an ambient background wash tinted from the app icon's palette. A new Layout menu lets you force Automatic, Sidebar, or Compact layout without resizing the window. Input & Output logging under 7-day and 30-day retention windows no longer silently drops new records, and the privacy dialog and settings copy now reflect your actual selected retention window.
v1.11.7
Interactive Dock Tile, session count fix, and model list scrolling
A new opt-in "Interactive Dock Tile" setting shows a live LED-style model marquee in the Dock while the proxy is running. The Home dashboard and menu bar no longer freeze the session request count at 500 on long-running sessions, the Proxy page's model list now scrolls correctly with large catalogs like OpenRouter's, upstream error messages carry more provider detail, and the Home dashboard's agent card now shows ProxyPilot Agent.
v1.11.3
I/O logging controls, export formats, and CLI session visibility
Input & Output logging retention now goes up to 30 days, with an optional external save location that falls back automatically if it becomes unreachable. Session History gains CSV and JSON file export. A new proxypilot sessions CLI command and MCP tool read session history and logs directly.
v1.11.2
Local and custom provider fixes, Copilot sidecar updates
Ollama, LM Studio, and custom OpenAI-compatible providers now reliably expose discovered models to Xcode. The GitHub Copilot sidecar gains version checking and one-click updates from Keys & Providers.
v1.11.1
Faster dashboard startup and cleaner reliability boundaries
Large retained sessions load with one dashboard update instead of hundreds, and model pricing metadata no longer requires repeated full-catalog scans. Custom-provider activation remains responsive, and test runs no longer reach production encrypted-log keys or leave launch tasks running after teardown.
v1.11.0
Managed ProxyPilot Agent for Xcode 27
Install, inspect, repair, or remove ProxyPilot Agent from the app or with proxypilot agent. The launcher can start the selected proxy route, keeps provider credentials in Keychain, and uses universal Apple Silicon and Intel helper binaries. Claude Agent routing stays intact for compatibility.
v1.9.6
9Router as a built-in local provider
Point ProxyPilot at a 9Router gateway and route Xcode Agent traffic through it. 9Router is treated as a local/helper provider on its local /v1 endpoint, needs no upstream provider key by default, and can store an optional endpoint token for gateways that require one.
v1.9.5
Custom providers you can actually select
Saved custom OpenAI-compatible providers now appear in the upstream picker, show an active state in Keys & Providers, persist across relaunch, and route with their own base URL and stored key. Custom-provider model lists and defaults stay separate from the built-in OpenAI provider.
v1.9.5
Session History stays smooth on large logs
Session History loads report-card metadata first and only decrypts prompt and output records for the session you open, instead of publishing every retained body into the UI on entry. Browsing long histories with large saved logs no longer stalls the screen.
v1.9.5
Report import off the main thread
Session-report import now runs as a background, fingerprint-gated task, so unchanged report files are skipped and decode work stays off the main actor. Refreshes feel lighter and resetting session stats cleanly cancels stale in-flight imports.
v1.9.1–1.9.3
Qwen joins the provider list
Qwen is now a direct cloud provider through Alibaba Cloud Model Studio's OpenAI-compatible DashScope route. The app keeps international and China-region key guidance separate, makes Qwen visible for existing customized provider lists, and labels the new route clearly in provider settings.
v1.9.0
Prompt caching is a first-class setting
ProxyPilot defaults supported routes to Auto cache mode, gives you Auto, Observe Only, and Off controls, and keeps cache accounting disabled when caching is off. Observe Only lets you see provider-reported cache behavior without changing outbound requests.
v1.8.9
DeepSeek session accounting
DeepSeek requests feed token and cost data into the shared session report surfaces, including cache-aware token counts when the provider returns them. Home and menu bar cost surfaces distinguish calculated provider cost from estimates.
v1.8.4–1.8.5
OpenRouter→Gemini schema hardening
Anthropic /v1/messages structured-output title generation routed through OpenRouter to Gemini keeps its schema constraint instead of returning Markdown-fenced JSON. MCP tool function parameter schemas containing enum on object or array nodes are cleaned for both direct Google routing and OpenRouter→Gemini routing, so Google's validator stops rejecting the request.
v1.8.3
OpenRouter cost counting + Fetch Live Models filter bar
Streaming requests now opt into stream_options.include_usage so OpenRouter and other OpenAI-compatible providers actually emit token counts in stream chunks and session cost can be computed. Fetch Live Models gets a search + provider + tier filter bar, which makes the ~263-item OpenRouter catalog usable instead of one long scroller.
v1.8.2
Audit fixes
Closes the first batch of confirmed audit issues from the v1.8.1 review: CLI daemon key exposure, unsafe JSON error fallback construction, and world-readable local log creation.
v1.8.1
Heartbeat telemetry and Copilot login hotfix
The app-open heartbeat stays minimal and version-focused, while broader diagnostics remain opt-in. Keys & Providers now guides unauthenticated Copilot users through login, distinguishes missing Copilot entitlement after GitHub auth, and offers a live sidecar check from inside the app.
v1.7.7–1.7.14
Agent reliability and security hardening
The GUI can now stop CLI-owned proxies. CLI session metrics feed into the GUI Session Report Card via a shared event store. Z.ai credential verification runs before agent launch — a stored-but-rejected key stops the session before opening Claude Code. Strict MCP argument validation returns structured errors for malformed types, out-of-range ports, and invalid filters. Missing-PID process discovery ensures correct status even without a PID file. Z.ai API key minimum length (20 chars) is enforced across GUI, CLI, and MCP auth paths. The built-in proxy rejects non-loopback clients before serving any response.
v1.7.4–1.7.6
Route visibility and evidence
A live Xcode-visible models panel refreshes /v1/models and shows the count and source. The Home dashboard now shows the running proxy's active model and flags picker changes as pending restart. A Last Xcode Agent Live Proof panel shows the most recent agent request model, status, and timestamp from the current session.
v1.7.3
Loopback security enforcement
The built-in GUI proxy rejects non-loopback client connections before parsing any request, so the proxy cannot answer /v1/models on the Mac's LAN address.
v1.6.5–1.6.7
Native Mac shell and customization hub
The app moved to a NavigationSplitView sidebar and native grouped toolbar with macOS 26 Liquid Glass on supporting systems and solid macOS 15-safe fallbacks. A Customization hub controls appearance, accent color, default section, Home section visibility, Launch at Login, and menu bar composition.
v1.6.0–1.6.3
Home dashboard and agent-first contracts
A full Home dashboard shows running state, active provider/model, request count, tokens, cost estimates, latency, issue status, recent requests, CSV export, and reset. The CLI/MCP contract gained schema-versioned JSON output, preflight and auth_status MCP tools, and Xcode 26.5 Locally Hosted copy for local provider setup.
v1.5.3–1.5.4
GitHub Copilot sidecar and OpenAI compatibility
GitHub Copilot provider routes through xcode-copilot-server with launchd/socket activation, install/remove controls, and explicit GitHub billing boundary copy. Direct OpenAI GPT-5.x and o-series token parameter compatibility ensures translated Xcode traffic reaches the new model family correctly.
How it works
Point Xcode at ProxyPilot. Choose the upstream yourself.
Xcode talks to a local proxy endpoint. ProxyPilot owns the routing layer: translating protocol shape, remapping model names, storing provider keys, checking setup, and keeping the selected upstream inspectable.
Xcode Agent
Uses Xcode Agent configuration to send model requests to the ProxyPilot endpoint on your Mac.
->
ProxyPilot
Runs on 127.0.0.1:4000, translates Anthropic and OpenAI-shaped traffic, remaps requested model names, tracks routing state, and records provider-reported usage details.
->
Your upstream
Route to z.ai, OpenRouter, OpenAI, Qwen, Gemini, Ollama/LM Studio, or your own OpenAI-compatible endpoint.
GUI surfaces
A Mac app built around the active session, not a pile of hidden settings.
Home dashboard
See running state, current provider/model, Xcode routing status, cache status, recent requests, tokens, latency, cost estimates, and CSV export from the same first screen.
Keys and providers
Each provider has its own Keychain-backed credential state. Local providers do not ask for cloud keys, Qwen includes region-aware key guidance, and custom providers can point to your own OpenAI-compatible endpoint.
Customization
Choose appearance, accent treatment, default window section, Home dashboard sections, Launch at Login, Liquid Glass control-strip behavior, and menu bar composition.
Automation
CLI and MCP are first-class, not afterthoughts.
ProxyPilot can run headless for terminal workflows, and the MCP server gives agents explicit recovery paths instead of forcing them to infer state from human strings.
Headless CLI
Use start, stop, status --json --require-running, auth, config, models --metadata, logs, launch, and update from scripts or agent workflows.
$ proxypilot status --json --require-running
$ proxypilot models --metadata --filter tool-calling
$ proxypilot serve --mcp15 MCP tools
Agents can preflight setup, check auth state, guardedly store secrets, verify local routing, start/stop/restart the proxy, install/remove Xcode config, list upstream models, read logs, and inspect session stats and session history.
Provider matrix
Cloud APIs, local servers, and custom routes.
ProxyPilot keeps provider choice explicit. It does not pretend every upstream behaves the same, and it keeps local/no-key routes separate from cloud-key flows.
| Route | Providers | Best fit |
|---|---|---|
| Cloud APIs | z.ai, OpenRouter, OpenAI, Google Gemini, xAI, Chutes, Groq, DeepSeek, Mistral, MiniMax, MiniMax CN, Qwen | Fast setup with your own provider keys and live model discovery. |
| Local gateways | Ollama, LM Studio, 9Router | No cloud API key. ProxyPilot detects localhost and avoids irrelevant key prompts. |
| Bring your own | Custom OpenAI-compatible endpoint | Private gateways, local labs, company proxies, or provider previews. |
Trust boundaries
The app says what it knows and keeps local checks local.
Routing status is scoped
GUI-owned proxy state, CLI/external listeners, cache reporting, and current-session metadata are labeled separately, so a single green light is not stretched into more certainty than it deserves.
Cache claims stay grounded
ProxyPilot shows provider-reported cache counters when they exist. It does not claim total savings unless the provider/model response gives enough information to support that claim.
Secrets stay out of files
Provider keys are stored via Keychain on macOS. File-backed fallbacks use restricted permissions, and logs/diagnostics redact bearer tokens.
FAQ
Useful answers first. Model-specific details one tab away.
The default view stays short for humans. The model-routing answers are still available when someone needs to confirm a specific upstream path.
Start here
The questions that remove the most friction.
Does ProxyPilot send my code anywhere by itself?
No. ProxyPilot forwards model requests only to the upstream provider you configure. The app runs locally on 127.0.0.1, and broader debugging analytics remain opt-in.
Which providers are supported?
ProxyPilot supports z.ai, OpenRouter, OpenAI, Google Gemini, xAI, Chutes, Groq, DeepSeek, Mistral, MiniMax, MiniMax CN, Qwen, 9Router, Ollama, LM Studio, and custom OpenAI-compatible endpoints.
Is ProxyPilot free?
Yes. ProxyPilot is free to download and use. You bring your own upstream provider keys, or use local providers such as Ollama and LM Studio without cloud keys.
Can agents control ProxyPilot from the terminal?
Yes. The CLI provides start, stop, status, auth, config, models, logs, launch, update, and MCP server commands for agent workflows.
Does ProxyPilot support Claude Agent and OpenAI-compatible providers?
Yes. ProxyPilot translates Anthropic-style agent traffic to OpenAI-compatible upstreams, including streaming and tool-call paths, while keeping provider credentials separate.
How do I change the LLM that Xcode Agent calls?
Point Xcode Agent at ProxyPilot's local endpoint. ProxyPilot then forwards the LLM call to the upstream you choose, including z.ai, OpenAI, Ollama, a custom OpenAI-compatible URL, or any other supported provider; Xcode itself still drives the agent the same way.
Specific upstreams
Common model families and how they route.
How do I use GLM with Xcode Agent?
Choose z.ai as the provider, add your z.ai API key, select the GLM model you want, and install the Xcode Agent configuration through ProxyPilot.
How do I use DeepSeek with Xcode Agent?
Choose a provider that hosts DeepSeek, such as Chutes or OpenRouter, add the provider key, and let ProxyPilot translate Xcode Agent requests to the selected OpenAI-compatible model route.
Can I use Qwen with ProxyPilot?
Yes. Qwen is available as a direct provider through Alibaba Cloud Model Studio's OpenAI-compatible DashScope route, and you can still use Qwen through OpenRouter or another compatible upstream.
How do I use Llama models?
Use Groq for fast hosted Llama inference, OpenRouter for broad model coverage, or any OpenAI-compatible endpoint that serves the Llama model you want.
Does ProxyPilot work with Mistral models?
Yes. Route through OpenRouter, Groq, Mistral-compatible gateways, or a custom OpenAI-compatible endpoint that exposes the Mistral model.
Can ProxyPilot route GPT-5.x or o-series models?
Yes. Choose OpenAI as the provider, add your OpenAI API key, and select the target GPT or o-series model. ProxyPilot handles the translated Xcode Agent request path.
How do I use Grok?
Choose xAI as the provider, add the xAI API key, select the Grok model, and install or refresh the Xcode Agent route.
Can I use Gemini models?
Yes. Choose Google Gemini directly or route Gemini through OpenRouter, then select the Gemini variant from the live model list.
Routes and boundaries
Where local, sidecar, and custom routes fit.
Which Xcode versions are supported?
ProxyPilot targets the current Xcode Agent configuration path and validates support during setup, instead of assuming every Xcode install exposes the same agent features.
Do local providers need API keys?
No. Ollama and LM Studio routes are local no-key flows. ProxyPilot detects localhost providers and avoids irrelevant cloud-key prompts.
What happened to the GitHub Copilot integration?
The experimental GitHub Copilot integration was removed in v1.15.1. ProxyPilot no longer installs, manages, or routes through its sidecar. Independently installed helpers are left untouched.
Can I use a custom endpoint?
Yes. Bring a private gateway, lab server, company proxy, or provider preview as long as it exposes an OpenAI-compatible endpoint.
Does ProxyPilot change how Xcode Agent builds, tests, or runs SwiftUI previews?
No. ProxyPilot only sits in the LLM request path. Xcode's native build, test, log, and preview integration with the agent continues to work exactly as Apple shipped it.
Can I bring my own API key (BYOK) for Xcode Agent?
Yes; that is the default flow. You add your own key for the upstream provider you want, or use Ollama or LM Studio locally with no key at all, and ProxyPilot routes Xcode Agent traffic through it.
Get ProxyPilot
Install the public app or the CLI.
Signed, notarized, and stapled with Gatekeeper verification. Sparkle auto-update is enabled for future releases.
macOS menu bar app
Settings UI, Xcode Agent setup, provider keys, Home dashboard, session stats, Sparkle updates, and menu bar controls.
$ curl -fsSL https://micah.chat/downloads/proxypilot-install.sh | bashHeadless CLI
Single binary for terminal workflows and MCP control. Use it when you want setup, proxy control, logs, and status without opening the GUI.
$ curl -fsSL https://micah.chat/downloads/proxypilot-cli-install.sh | bash