Paseo Dev Stack
AdvancedRun Claude Code, Codex, and other coding agents on your own machine with Paseo and steer them from any device.
Published 27 September 2026
About Paseo Dev Stack
Paseo is a self-hosted, open-source control plane for coding agents. A daemon runs on your laptop, a home server, or a VPS and manages the official agent CLIs unmodified, with that machine's full dev environment and your existing subscriptions. Every client connects to the same daemon: desktop apps for macOS, Windows, and Linux, a web app, iOS and Android apps, and a CLI, so a session started at your desk picks up on your phone without losing state.
In this stack agents and models are separate choices, and both allow more than one pick. Claude Code is the default agent, with Codex, OpenCode, Pi, and Oh My Pi alongside it, and more through the Agent Client Protocol. Each agent has a home model family, but the model layer isn't welded to the agent: OpenCode, Pi, and Oh My Pi run almost any provider, and Codex can be pointed at other providers too. Optional git worktrees give parallel agents their own branches.
Paseo is built for checking in from anywhere. Push notifications tell you when an agent needs input or finishes, and voice mode lets you steer a session by talking, with speech processed on your device by default. Paseo is free under the Apache 2.0 license, relay included, with no telemetry and no account required.
The daemon usually starts on your own computer, since the desktop app bundles one, and its agents pause whenever that computer sleeps. Every Paseo client talks to the daemon the same way, though, so it can move to a Mac mini, a home server, or a VPS with no change to the apps you use: desktop, web, and phone reach it through Paseo's free end-to-end encrypted relay with no open ports, over Tailscale or another VPN if you already run one, or through an SSH tunnel. That always-on setup is the one Paseo is designed around. CI/CD is a common addition once agents open pull requests on their own.
Key Features
- ✓Self-hosted daemon that runs official agent CLIs unmodified in your own dev environment
- ✓Pick one or more agents: Claude Code, Codex, OpenCode, Pi, and Oh My Pi, with more through ACP
- ✓Pick one or more models, with OpenCode, Pi, and Oh My Pi opening the layer to any provider
- ✓Desktop (macOS, Windows, Linux), web, iOS, Android, and CLI clients on the same sessions
- ✓The same apps work whether the daemon runs locally or on a server
- ✓Push notifications when an agent needs input or finishes a step
- ✓Free end-to-end encrypted relay, Tailscale or VPN, or SSH tunnel for remote access
When to Use Paseo Dev Stack
- →Running agents on a home server or VPS and steering them from a phone
- →Starting a task at the desk and answering the agent's questions later from anywhere
- →Keeping every agent session and all code on hardware you control, with no vendor account
- →Driving an agent by voice while away from a keyboard
- →Mixing premium and open-weight models across several agents on one always-on machine
Pros
- Free, Apache 2.0-licensed, fully self-hosted, with no telemetry or required account
- Agents run unmodified, so each behaves exactly as it does in its own terminal
- True cross-device continuity: the same session on desktop, web, phone, and CLI
- Flexible remote access, from a zero-setup encrypted relay to your own Tailscale network
Cons
- You run and maintain the daemon yourself, more setup than a desktop-only app
- A smaller community and less documentation than the leading agent orchestrators
- No hosted cloud, so agents running with the laptop closed need a server of your own
- Every agent needs its own subscription or API key, so costs add up across providers
Coding Agent Options for Paseo Dev Stack
Paseo Dev Stack with Claude Code
Claude Code is the default agent here: Anthropic's CLI, strong at multi-file changes and a natural pick for the hardest task on the daemon. It bills through a Claude plan from $20/month or per token via the API. It's built for Claude models; DeepSeek and GLM publish Anthropic-compatible endpoints it can point at, outside Anthropic's support.
Paseo Dev Stack with Codex
OpenAI's agent CLI, included with every ChatGPT plan down to the free tier. A Codex session next to Claude Code is the simplest way to compare two vendors on the same task. It defaults to OpenAI's models but officially supports other providers through its config, including OpenRouter and local models.
Paseo Dev Stack with OpenCode
Open-source and model-agnostic, OpenCode lets a session run Gemini, DeepSeek, Kimi, a local model, or nearly any other provider with an API key, billed per token by that provider. The usual pick for long-running bulk work on a server, where a cheap model keeps costs down.
Paseo Dev Stack with Pi
Pi is a deliberately minimal, model-agnostic agent harness shaped through TypeScript extensions, with any provider's key behind it. Its small footprint suits long-running sessions on a modest server, and Paseo supports it natively.
Paseo Dev Stack with Oh My Pi
A feature-heavy fork of Pi with 31 built-in tools, LSP integration, browser automation, and 60+ model providers, for developers who want Pi's openness with far more built in. Paseo supports it natively, alongside Pi.
These are highlighted picks. To see all the tools, check the AI Coding Agents category.
LLM Options for Paseo Dev Stack
Paseo Dev Stack powered by Claude
Claude is the default model, the one Claude Code is built for, and a common choice for the session doing the most careful reasoning. It's also available in OpenCode, Pi, and Oh My Pi, so one Claude key or plan can back more than one session at once.
Paseo Dev Stack powered by OpenAI
OpenAI's GPT models are Codex's home family and are also reachable through OpenCode, Pi, and Oh My Pi. Picking them alongside Claude gives a second vendor's approach to the same problem, and a ChatGPT plan covers Codex usage without per-token billing.
Paseo Dev Stack powered by Google Gemini
Gemini's long context window suits repository-wide questions and large refactors, and its Flash models are fast and cheap for routine work. In this stack it's reached through OpenCode, Pi, or Oh My Pi.
Paseo Dev Stack powered by DeepSeek
An open-weight model priced far below the premium providers per token, a good match for agents left running on a server for high-volume work like test generation. It runs through OpenCode, Pi, or Oh My Pi, or behind Claude Code via DeepSeek's Anthropic-compatible endpoint.
Paseo Dev Stack powered by Kimi
Moonshot's open-weight Kimi models are another budget option tuned for agentic coding, with long context for large repositories. Use them through OpenCode, Pi, or Oh My Pi.
These are highlighted picks. To see all the tools, check the LLM category.
Version Control Options for Paseo Dev Stack
Paseo Dev Stack with GitHub
GitHub is the default here, with the largest pull-request ecosystem and the review bots and CI tools most agent workflows assume. Paseo itself works the same with any git host.
Paseo Dev Stack with GitLab
GitLab fits when source control has to be self-hosted or stay on your own infrastructure. Paseo only sees git, so agents push to GitLab exactly as they would to GitHub.
Paseo Dev Stack Add-ons
Each addition below extends this stack with a capability the base stack works fine without. None are required: include the ones your product actually needs when building this stack, and skip the rest.
Server Add-ons
Add a server when you want the agent to keep running somewhere other than your own laptop, reachable at any time and not tied to your machine staying on. Some agents also offer their own managed cloud sessions as an alternative to self-hosting; check the stack's own description for details.
Paseo Dev Stack with Hetzner
Hetzner is the default here: a 4 GB server, Claude Code's minimum for one agent, is about €5.49/mo, and 8 GB or more suits several sessions for little extra. It's a plain Linux VPS for the Paseo daemon, with data centers in Germany, Finland, the US, and Singapore to keep the relay or Tailscale connection close.
Paseo Dev Stack with DigitalOcean
DigitalOcean's 4 GB Droplet is $24/mo and 8 GB $48/mo, more per gigabyte than Hetzner, with more regions and one-click snapshots and backups. Its 1-Click Codex CLI and OpenCode Droplets come with two of Paseo's agents installed; the $4 plan's 512 MB is too small for any agent.
Paseo Dev Stack with Hostinger
Hostinger's KVM 1 has 4 GB of RAM at $6.49/mo introductory, renewing at $11.99/mo, enough for the daemon and one agent, and KVM 2 doubles the memory for more sessions. Its Claude Code VPS template preinstalls Paseo's default agent, so only the daemon is left to add.
Paseo Dev Stack with Raspberry Pi
A Raspberry Pi 5 with 4–16 GB of RAM makes a home-server host for a one-time cost instead of a monthly bill, the home-lab setup Paseo is built for, reached through Paseo's relay without opening ports. Memory caps the number of agents: 8 GB suits two or three, 16 GB a few more.
These are highlighted picks. To see all the tools, check the Hosting & Cloud category.
Remote Access Add-ons
Add remote access when you want to reach an agent running on another machine — a VPS or a home server — without exposing it to anyone but you.
Paseo Dev Stack with Tailscale
Add Tailscale when you'd rather reach the Paseo daemon over your own private network than through Paseo's relay. Nothing is exposed to the public internet, and the personal tier is free.
Code Review Add-ons
Add code review when you want an AI reading every pull request before it merges: it comments on the diff so bugs, security issues, and inconsistencies surface before a human has to catch them.
Paseo Dev Stack with CodeRabbit
Add CodeRabbit when several agents are opening pull requests faster than one person can read them: it reviews every GitHub or GitLab diff with a summary and line-level comments, with a free tier for open-source repos.
Paseo Dev Stack with Greptile
Add Greptile when parallel agents produce more diffs than one person can review: it indexes the whole repository so its comments carry context beyond the changed files, and it can be self-hosted for full control.
Paseo Dev Stack with Cursor Bugbot
Add Cursor Bugbot when a bug-focused reviewer is enough: it reviews GitHub or GitLab pull requests whatever wrote the code, though it bills through a Cursor account and plan rather than its own subscription.
CI/CD Add-ons
Add CI/CD when you want a dedicated pipeline for running tests, linting, or multi-stage builds before a deploy goes out. Many hosting platforms already redeploy automatically on every push on their own — a CI/CD tool adds the most value on top of that by gating the deploy on a passing test suite, and matters even more when the hosting choice does not auto-deploy at all, such as a self-hosted server.
Paseo Dev Stack with GitHub Actions
Add GitHub Actions when several agents are opening pull requests at once and every branch should be tested automatically before you spend time reviewing it.
Paseo Dev Stack with GitLab CI/CD
Add GitLab CI/CD for the same automatic testing when the repo lives on GitLab, with pipelines defined in a .gitlab-ci.yml file that can run on your own runners.
These are highlighted picks. To see all the tools, check the CI/CD Pipelines category.
Frequently Asked Questions about Paseo Dev Stack
Relay, Tailscale, or direct connection: which should I use?
The relay is the easiest: the daemon connects out to Paseo's hosted relay and your devices meet it there, end-to-end encrypted, with no open ports and no network setup. Tailscale or another VPN suits anyone already running one, or who wants no third party in the path. An SSH tunnel through your own OpenSSH client works too. Exposing the daemon port directly only makes sense on a trusted local network.
Where should the Paseo daemon run?
Wherever the agents' code and tools live, since agents work in that machine's environment. A laptop works but pauses every session when it sleeps. A home server or VPS keeps agents running around the clock, and it doesn't cost you the interface: the desktop, web, and phone apps connect to a remote daemon exactly as they do to a local one. Size it by agent count: 4 GB covers Claude Code alone, and several agents working at once need more.
Which agents can run which models?
Most combinations work, with varying effort. OpenCode, Pi, and Oh My Pi take almost any provider's key. Codex defaults to OpenAI but officially supports other providers through its configuration, local models included. Claude Code is built for Claude, and Anthropic supports only Claude models, though DeepSeek and GLM publish Anthropic-compatible endpoints some developers point it at.
What is ACP, and what does it add?
The Agent Client Protocol is an open standard, started by Zed, for connecting coding agents to the apps that drive them. Paseo speaks it, so agents that implement ACP can run on the daemon beyond the ones Paseo supports by name.
How is Paseo different from Orca or T3 Code?
Orca and T3 Code are desktop-first apps that can also connect to a server. Paseo starts from the other end: a headless daemon on a machine you choose, with every client, desktop included, as a remote control for it. That makes it the most natural fit for agents on a server, and the least polished as a local desktop app.
Stacks Related to Paseo Dev Stack
Claude Code Dev Stack
DeveloperClaude Code in the terminal, desktop app, or web, paired with Claude and the add-ons to run it anywhere.
T3 Code Dev Stack
Developer
One open-source app for Claude Code, Codex, OpenCode, and other coding agents, on desktop and mobile.
Conductor Dev Stack
Developer
Run Claude Code, Codex, and Cursor's agent in parallel on your Mac with Conductor, each in its own workspace.
Orca Dev Stack
Developer
Run Claude Code, Codex, and other coding agents side by side in Orca, each on its own branch and model.
Scores
Popularity2/5
Paseo is a small open-source project in a young category, known mostly among developers already self-hosting their agents.
Learning Curve3/5
Installing and running a daemon, choosing how clients reach it, and setting up the server's dev environment come before the first agent session, though the clients themselves are simple.
Flexibility5/5
Five named agents plus any ACP agent, a wide-open model layer, optional worktrees, and a daemon that runs on a laptop, home server, or VPS with three ways to connect.
Performance3/5
The daemon is light, and agents run with the full speed of the machine hosting them, but responsiveness from remote clients depends on the connection path.
Portability5/5
Apache 2.0-licensed and fully self-hosted, with no account or telemetry, and every agent, model, and git host underneath can be swapped.
Tools in the Paseo Dev Stack Stack
Development Tools
Add-ons (optional — add any, or none)
Server
Remote Access
Code Review
CI/CD
Paseo Dev Stack Pricing
Paseo is free and open source, relay included, so the bill is the agents and, optionally, a server. Claude Code starts at $20/month with no free tier, Codex has a limited free tier with Plus at $20/month, and both reach $200/month; OpenCode, Pi, and Oh My Pi are free but bill per token through the provider you connect. An always-on VPS is about €5.49/month for 4 GB and more for several agents; a Raspberry Pi is a one-time cost instead.
Apache 2.0-licensed daemon and clients, including the end-to-end encrypted relay.
Claude Code from $20/mo (no free tier); Codex free tier, Plus $20/mo; both up to $200/mo. One subscription per agent.
Model-agnostic agents bill per token with the provider you connect; open-weight models like DeepSeek and Kimi cost a fraction of premium ones.
Free tier covers individual devs; GitHub Team is $4/mo, GitLab Premium is $29/mo for more seats.
Tailscale's personal tier is free for up to 100 devices.
4 GB runs one agent, 8 GB or more suits several: Hetzner from €5.49, Hostinger $6.49 intro ($11.99 renewal), DigitalOcean $24–48; or a Raspberry Pi as a one-time cost.