# Install magpie with Homebrew Cask Model gateway for AI coding agents. Version 0.1.2900 via Homebrew Cask; verified from local package data. ## Install ```sh sudo av install cask:magpie ``` Additional install commands: ### macOS - Homebrew Cask (100%): ```sh brew install --cask magpie ``` Evidence: local Homebrew cask metadata ## Package facts - **Package key:** cask:magpie - **Package manager:** Homebrew Cask - **Package manager page:** - **Version:** 0.1.2900 - **Source summary:** Model gateway for AI coding agents - **Homepage:** - **Repository:** - **Upstream docs:** - **Source archive:** - **Generated:** 2026-11-04T12:20:42+00:00 ## Install behavior - Post-install hook: not defined - Bottle: not available ## Freshness - Page generated: 2026-11-04 - Package-manager version: 0.1.2900 - Local data: ok - Upstream repository: https://github.com/yetone/magpie-releases - info: No package-manager update timestamp was available. - info: No cached GitHub release or tag data was available. ## From the blog - [Who should choose the model?](https://pkg.so/blog/2026-11-01/who-chooses-the-model/) - [A Local Control Plane for Coding Agents](https://pkg.so/blog/2026-10-28/a-local-control-plane-for-coding-agents/) ## Project history and usage magpie is a local model gateway and configuration hub that lets multiple AI coding agents use models from subscriptions, API vendors, compatible endpoints, and local runtimes. ### Project history magpie was developed as an open-source local model gateway and configuration application for AI coding agents. The project combines a desktop interface, terminal operation, and a loopback gateway listening by default on 127.0.0.1:3425. ### Adoption history magpie supports macOS, Windows, and Linux and recognizes a growing range of coding agents, including Claude Code, Codex, Gemini CLI, OpenCode, Goose, Cursor CLI, and Copilot CLI. It also integrates subscription accounts, API-key providers, compatible endpoints, Ollama, and LM Studio. ### How it is used Users add subscription accounts, provider API keys, or local endpoints once, then select a model or routing group for each detected coding agent. magpie updates that agent's native configuration and serves requests through its local gateway. The application can remain in the tray, run only the gateway, or install a launchd or systemd service. New agent sessions pick up model changes after their configuration is rewritten. ### Why package nerds care magpie addresses the increasingly fragmented configuration layer around AI coding agents. Instead of configuring each agent independently for every model provider, users maintain providers and routing centrally while magpie writes narrowly scoped entries into each agent's native configuration. Its local OpenAI-, Anthropic-, and Gemini-compatible gateway makes otherwise incompatible agents and model providers interoperable, while routing groups support multiple accounts, fallback, quota-aware selection, and intent-based routing. ### Timeline - Initial direction: magpie unified model selection and provider setup for coding agents behind a local gateway. - Current project: The application supports desktop, terminal, tray, daemon, and browser-oriented operation across macOS, Windows, and Linux. ### Related projects - Claude Code, Codex, Gemini CLI, OpenCode, and other coding agents can be configured to send requests through magpie. - Ollama and LM Studio can supply locally hosted models to magpie without API keys. ### Sources - - - - ## Security Notes No matching local secret-handling manifest was found for magpie. Package metadata is still published here so future coverage has a stable package URL. ## Configuration and credential file locations These source-backed paths show where this package keeps local settings or durable credentials. Automic Vault can use them as review targets for secret scanning, migration, and command approval. ## Configuration files - Unix: ~/.config/magpie/settings.json - Windows: ~/.config/magpie/settings.json ## Credential files - Unix: ~/.config/magpie/providers.json - Windows: ~/.config/magpie/providers.json ## Source Database Details - **Source Database:** Homebrew cask API - **Tap:** homebrew/cask - **Full Token:** magpie - **Names:** magpie - **Depends On:** Macos: >=: 12 - **Artifacts:** App: 1, Target: 1, Zap: 1 - **Auto Updates:** yes - **Deprecated:** no - **Disabled:** no ## Related links - [Homebrew Cask packages](https://pkg.so/homebrew-cask-packages/) - Generated from Homebrew Cask package-manager metadata. - [Homebrew Cask applications](https://pkg.so/homebrew-cask-applications/) - Cask artifact metadata includes an app bundle. - [ollama](https://pkg.so/brew/ollama/) - Use Ollama to run open models locally, then add it as an on-machine provider in magpie so coding agents can reach those models through magpie's gateway. Sources: · - [lm-studio](https://pkg.so/cask/lm-studio/) - LM Studio can serve downloaded local models through compatible APIs, while magpie exposes those models to its configured coding agents as an on-machine provider. Sources: · ## Combined YAML source View the package source record on GitHub. [combined/magpie.yml](https://github.com/mxcl/pkgdb/blob/main/combined/magpie.yml) ## Sources - package-page enrichment - curated configuration and credential file locations - curated package history - package version freshness - package relationship graph - cross-ecosystem install command graph