ClaudeMap

Agent Launcher

One desktop app to configure and run the coding-agent CLIs you already have — Claude Code, Codex, Gemini CLI and more — without touching their internals.

SDKs & Toolingdesktop-appcli-launcherprofilesmcpmulti-agent
Visit projectagent-launch/agent-launcher

A local desktop workspace that detects the coding-agent CLIs already installed on your system, links them (never reinstalling or updating a CLI you have), and installs missing ones in one click through their official installers — Claude Code via its own installer with an npm fallback. Six agents are supported: Claude Code, Codex CLI, OpenCode, Pi, Gemini CLI and Hermes Agent. Each agent keeps multiple provider profiles; when a profile changes, the app synchronizes the environment variables and native config files that CLI expects (Claude Code settings, Codex config.toml and auth.json, OpenCode opencode.json and more), and a real minimal model request checks endpoint, credentials, model and account status before the profile is used. Sessions run in an embedded terminal or chat view, launched in a chosen project folder and resumable from the working directory the CLI recorded. The app reads each CLI's own local conversation history (JSONL and SQLite layouts), can inspect installed MCP servers and skills, and summarizes token counts, request counts, sessions and models in a local usage dashboard. Available for macOS (DMG and ZIP for Intel and Apple silicon), Windows (NSIS) and Linux (AppImage).

What it does

  • Links existing CLIs instead of owning them — no silent replacements, no global npm state changes, no CLI redirected into an app-specific config home
  • Per-agent provider profiles synced into each CLI's native config, with a real minimal model request as a pre-flight connectivity check
  • Project-aware sessions, local conversation history browsing (JSONL/SQLite), MCP server and skills inspection, and a local usage dashboard with optional pricing data
  • CI builds, signed-and-verified artifacts when signing secrets are configured; MIT with bilingual EN/ZH docs

When to use it

Multi-agent users switching between Claude Code, Codex and Gemini CLI in one place — one profile switch instead of re-editing three config formats; also a quick path to a working setup on a new machine.

Getting started

# download from GitHub Releases:
#   macOS DMG/ZIP (Intel + Apple silicon) · Windows NSIS · Linux AppImage
https://github.com/agent-launch/agent-launcher/releases

# or from source (Node 22+, pnpm):
pnpm install && pnpm build

Things to know

MIT. API keys are deliberately stored as plaintext in ~/.agent-launcher/config.json (and CLI-native config files when required) — masked in the UI but readable by your local user account, the README says so openly. Without signing credentials the builds publish unsigned, so expect SmartScreen/Gatekeeper warnings; Linux AppImage may need FUSE 2. Local-first, not offline-only: running an agent still talks to the selected provider, and version checks may contact npm, PyPI or GitHub.

Repository

agent-launch/agent-launcher

Charted

Related on the map

  • Harness Bridge

    SDK / Tool

    A local launcher that points Claude Code and seven other coding harnesses at any OpenAI-, Anthropic- or Responses-compatible endpoint — without editing a single config file.

    • launcher
    • local-models
    • cli
    • menu-bar
    View

  • Claude for Siri

    SDK / Tool

    Answers macOS Spotlight and Siri prompts with your signed-in Claude Code account — an experimental macOS 27 model-delegation provider.

    • macos
    • siri
    • spotlight
    • model-delegation
    View

  • AI Coding Assistant Data Extractor

    SDK / Tool

    Extract your own local chat history from ten AI coding assistants into one normalized JSONL — for fine-tuning, analytics, or backing up years of conversations. Standard library only.

    • cli
    • data-export
    • jsonl
    • backup
    View