fleet

Every git repo you own, at a glance.

fleet is a desktop dashboard for every git repository under your project roots. See which are dirty, behind, or stale - then fetch, commit, resolve conflicts, cherry-pick, search across all of them, and track tasks and deadlines, from one window. No terminal required.

Download  ·  Landing page  ·  Changelog

Why fleet

Most git GUIs open one repository at a time. fleet is built for the whole fleet of them: dozens of repos across your project roots, monitored like a status board.

• Live status - dirty / behind / ahead / stale / CI-failing across every repo; fetch, pull or push in bulk.
• Git without a terminal - per-file staging, commit, amend, merge/rebase a diverged upstream, cherry-pick, reflog recovery, conflicts resolved inline.
• Project management - tasks, deadlines, tags, a cross-project Agenda and a Today view.
• AI brief & agent - a brief that weighs GitHub CI/PR/issue signals, per-repo chat, and an agent that runs under per-action approval (defaults to the claude CLI; OpenAI/Gemini optional).
• Search - fuzzy file quick-open and case-insensitive content search across every repo, plus a command palette.
• Optional sync - sign in with GitHub to sync projects and intel across devices; export/import everything as JSON.

Run

Download the binary for your platform from Releases - fleet.exe (Windows), fleet-macos.zip (macOS), fleet (Linux) - or build from source:

wails build
./build/bin/fleet.exe

First run writes a config and creates the data directory next to it:

OS	Path
Windows	%APPDATA%\fleet\config.toml
macOS / Linux	$XDG_CONFIG_HOME/fleet/config.toml, else ~/.config/fleet/config.toml

roots defaults to ~/Projects. Edit it and restart.

Build from source

Requires Go 1.22+, Node 18+, and Wails v2 (go install github.com/wailsapp/wails/v2/cmd/wails@latest). On Linux also install libgtk-3-dev and libwebkit2gtk-4.1-dev.

wails build            # desktop app -> build/bin/fleet.exe

The AI brief, repo chat and agent features shell out to the claude CLI, which must be on PATH (it is the default AI provider). The OpenAI and Gemini providers are configured in Settings instead and need no CLI.

The landing page in site/ is plain HTML deployed by .github/workflows/pages.yml. Its favicon, touch icon and Open Graph card are generated from build/appicon.png by python tools/make-site-icons.py (Pillow only); re-run it if the app icon changes.

Config

roots = ["C:/Users/you/Projects"]
scan_depth = 2
editor = "code"
terminal = "wt"
auto_fetch_minutes = 0
show_non_git = true

Backend server (fleetd)

cmd/fleetd is the fleet cloud backend: GitHub OAuth identity plus a per-user versioned document store with Last-Write-Wins sync. It is deployed to Fly.io (region nrt) on Neon serverless Postgres. The desktop app never imports it.

Environment

Variable	Purpose
DATABASE_URL	Neon Postgres connection string (postgres://...?sslmode=require)
JWT_SIGNING_KEY	HS256 secret for access tokens. At least 32 bytes with real entropy - the server refuses to boot otherwise. Generate with openssl rand -base64 48
GITHUB_OAUTH_CLIENT_ID	GitHub OAuth app client id
GITHUB_OAUTH_CLIENT_SECRET	GitHub OAuth app client secret
GITHUB_OAUTH_CALLBACK_URL	This server's public callback, e.g. https://fleet-api.fly.dev/auth/github/callback
FLEET_ALLOWED_REDIRECT	Allowed loopback redirect prefix (default http://127.0.0.1)
PORT	Listen port (default 8080)
TRUST_PROXY	Trust Fly-Client-IP/X-Forwarded-For for the client IP (default false)

TRUST_PROXY is required on Fly.io. Fly terminates TLS and proxies every request to the app, so without TRUST_PROXY=true the server sees only Fly's proxy address as the client IP and the per-IP rate limiter on /auth collapses onto a single bucket for all users. It is set in fly.toml's [env] block, alongside PORT, since it is a deploy-topology flag rather than a secret.

Neon

1. Create a Neon project and open its dashboard's "Connection Details".
2. Copy the pooled connection string (the one using the PgBouncer host, not the direct host) and make sure sslmode=require is on it, e.g. postgres://user:pass@ep-xxx-pooler.region.aws.neon.tech/fleet?sslmode=require. Use this as DATABASE_URL.
3. Migrations (golang-migrate, embedded) run automatically at startup.

GitHub OAuth app

1. Create an OAuth app at https://github.com/settings/developers (or in an org's settings) for fleet-api.
2. Set the Authorization callback URL to https://fleet-api.fly.dev/auth/github/callback (substitute your app's <app>.fly.dev hostname if different from fleet-api). This must match GITHUB_OAUTH_CALLBACK_URL exactly.
3. The server requests scopes read:user user:email during the OAuth handshake; no other scopes are needed.
4. Copy the generated client id/secret into GITHUB_OAUTH_CLIENT_ID / GITHUB_OAUTH_CLIENT_SECRET (set as Fly secrets, below).

Fly.io secrets

fly apps create fleet-api
fly secrets set \
  DATABASE_URL="postgres://user:pass@ep-xxx-pooler.region.aws.neon.tech/fleet?sslmode=require" \
  JWT_SIGNING_KEY="$(openssl rand -base64 48)" \
  GITHUB_OAUTH_CLIENT_ID="..." \
  GITHUB_OAUTH_CLIENT_SECRET="..." \
  GITHUB_OAUTH_CALLBACK_URL="https://fleet-api.fly.dev/auth/github/callback"
fly deploy

PORT and TRUST_PROXY are non-secret and already set in fly.toml's [env] block, so they do not need fly secrets set. Set the GitHub OAuth app's Authorization callback URL to https://fleet-api.fly.dev/auth/github/callback (see above). fly.toml health-checks GET /healthz on internal_port = 8080, matching the server's default PORT; response body is ok.

Run tests locally

# Unit/handler tests (no database needed):
go test ./cmd/fleetd/... ./internal/server/...
# Postgres repo tests (start a throwaway Postgres, then):
DATABASE_URL_TEST="postgres://postgres:postgres@localhost:5432/fleet_test?sslmode=disable" \
  go test ./internal/server/pgstore/...

Local dev/e2e: docker compose up -d postgres then go test ./... with DATABASE_URL_TEST=... (runs the pgstore and internal/server/e2e suites against it); docker compose up -d --build fleetd runs the server via the repo's docker-compose.yml.

License

Copyright (c) 2026 H.K. All rights reserved.

fleet is source-available, not open source. It is licensed under the PolyForm Noncommercial License 1.0.0 (see LICENSE): you may use, study, modify and share it for any noncommercial purpose - personal use, research, education, hobby projects, and nonprofit/government organizations. Commercial use is not permitted under this license.

The copyright holder reserves all commercial rights. A commercial license may be offered separately in the future; contact the author for commercial use.

Third-party dependencies keep their own (permissive / MPL-2.0) licenses and are unaffected by fleet's license.