MCP server de devstack
Commit 14 nov 2026Permite que los agentes de programación gestionen, monitoreen y desplieguen stacks de desarrollo en máquinas locales y grupos de servidores remotos.
También conocido como nuetzliches/devstack
- Stars
- 5
- Herramientas
- —
- Descargas, 30 días
- —
- Última versión
- —
- Último commit
- 14 nov 2026
- Licencia
- Apache-2.0
¿Qué es devstack MCP?
devstack es un MCP server. Permite que los agentes de programación gestionen, monitoreen y desplieguen stacks de desarrollo en máquinas locales y grupos de servidores remotos.
Resumen generado a partir del README del proyecto
Acerca del MCP server de devstack
Los agentes de programación pueden gestionar, monitorear y desplegar stacks de desarrollo completos en máquinas locales y grupos de servidores remotos usando devstack. Este MCP server permite que los clientes de IA interactúen con tu infraestructura a través de stdio, proporcionando herramientas para iniciar, detener, reiniciar y solucionar problemas en los servicios. Está diseñado para garantizar que los cambios solo se implementen cuando sean estables, evitando que builds rotos afecten a tu entorno de ejecución.
Los desarrolladores utilizan esto para ejecutar lanes aislados para cada git worktree, asegurando que múltiples agentes puedan trabajar en diferentes ramas sin colisiones de puertos. Cuando los recursos locales se agotan, el sistema puede mover estos lanes a un pool remoto de máquinas, redirigiendo los puertos de vuelta a tu configuración local. La configuración se basa en archivos Compose para la infraestructura y un archivo devstack.toml para las definiciones de los servicios. Funciona de forma nativa en Windows y Linux, con macOS disponible en versión preview.
Al instalar skills directamente en tu repositorio, los agentes obtienen la capacidad de comprobar si un cambio está activo, ver logs y realizar rollbacks. Esto crea un ciclo cerrado donde una IA puede intentar una corrección, verificar el estado del build y promover el servicio una vez que supere las comprobaciones de salud.
Qué puedes hacer con él
- 1Verificar la salud del servicio tras los cambiosPide a un agente que compruebe si un cambio de código específico se ha implementado correctamente y está activo actualmente.
- 2Delegar builds pesados a servidores remotosIndica a tu agente de programación que mueva el stack de desarrollo actual a una máquina en el pool de servidores del equipo.
Escrito a partir del README del proyecto. Cada caso de uso se verifica con una cita del mismo.
Verificaciones automáticas
Ejecutado 15 nov 2026Informe completo →Aún no verificado. Un servidor necesita una puntuación de 80, ningún hallazgo grave y suficientes datos para que podamos revisarlo. Cómo funcionan las verificaciones · Qué corregir para obtener más puntos
- Repositorio de origen vinculado y accesiblenuetzliches/devstack is public and reachablegithub.com/nuetzliches/devstack ↗Aprobado
- El paquete indica este repositorio como su origenNo verificado
- Procedencia de la compilación (atestación de Sigstore)No verificado
- Licencia de código abierto (OSI)Apache-2.0spdx.org/licenses/Apache-2.0.html ↗Aprobado
- Commit o lanzamiento en el último añoLast commit or release 2026-11-14github.com/nuetzliches/devstack ↗Aprobado
- OpenSSF ScorecardNo verificado
- El nombre del paquete no es similar a un paquete popularNo verificado
- No hay avisos críticos en el paquete ni en sus dependenciasNo verificado
- Sin instrucciones ocultas para el modelo0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersAprobado
- Las herramientas destructivas están etiquetadasNo verificado
- El endpoint remoto responde al handshake de MCPNo verificado
README
Ver en GitHubdevstack
Your whole stack, always running the code you saved, never a broken build. Native services, their containers, every git worktree beside it, and when your laptop runs out of room, the team's server: one CLI, one dashboard, one MCP server for your coding agents.
devstack up # infrastructure, builds, every service: healthy or it says why
devstack ui # every stack, live, in the browser
- Save, and it is live, or it says why not. A change builds as a
candidate beside the running instance and takes over only once it is
healthy. A compiler error leaves the old build answering;
devstack statussaysnot yet active: building | failed | helduntil your change runs. - Every worktree is a lane.
git worktree addanddevstack upin it: the same stack on its own ports, with its own containers and data, beside the main one. Ten agents on ten branches, no collisions. - Lanes on other machines.
devstack up --on build-serverruns a worktree's lane on a machine of the team's pool, its ports forwarded to yours as if it ran locally. Move it back with one command, share it with a reviewer, give it the main ports. - Made for agents. An MCP server and generated skills let coding agents start, check, fix and move the stack, and know whether their change is live.
- Your files, no lock-in. Infrastructure comes from the Compose files you
deploy with; services from a short
devstack.toml.devstack env api --export dotenvruns any service without devstack. - Windows and Linux, natively. One static binary, no WSL required. macOS is in preview.
Before 1.0 · local stacks, lanes and the pool are in daily use on .NET, Node and Angular stacks. See what is planned and the decisions behind it.
Install
# Windows: to %LOCALAPPDATA%\Programs\devstack, on PATH
irm https://github.com/nuetzliches/devstack/releases/latest/download/install.ps1 | iex
# Linux (x86_64, arm64; static) and macOS: to ~/.local/bin
curl -fsSL https://github.com/nuetzliches/devstack/releases/latest/download/install.sh | sh
# macOS and Linux, with Homebrew
brew install nuetzliches/tap/devstack
On Windows, winget install nuetzliches.devstack installs the same build
(releases after 0.15.0, once winget's catalogue has them, usually a few days
after the release); devstack update then updates it through winget, and
the first devstack up registers the hub as the installer would. A copy
Homebrew installed is updated the same way, through brew upgrade.
macOS is a preview (ADR-0054, what a tester checks):
Apple Silicon and Intel, macOS 11 or later, Docker Desktop, OrbStack or
Colima. None of the maintainers has a Mac: CI checks what it can, and
reports check the rest.
Desktop notifications, the devstack:// link and a Mac as a pool node are
not there yet. The binary is not signed: install it with Homebrew or
install.sh (neither marks it for Gatekeeper); a copy downloaded with a
browser is blocked until xattr -d com.apple.quarantine devstack.
The scripts and devstack update check the download against the release's
SHA256SUMS: that proves the archive is the one the release lists (not
damaged or swapped on the way), not who built it, since both come from the
same release. Each release also carries a build provenance attestation; with
the GitHub CLI installed and logged in, the scripts
verify it (gh attestation verify <archive> --repo nuetzliches/devstack):
built by this repository's release workflow (releases from 0.10.0 on). The
Windows executable is not code-signed, so SmartScreen or an endpoint
protection product may warn about it at first (SECURITY.md). The
landing page tells whether your
machine is ready. devstack update (or the dashboard) installs a newer
release; devstack self uninstall removes devstack again. From source: cargo install --git https://github.com/nuetzliches/devstack devstack (without the
dashboard: it needs npm ci && npm run build in ui/ before the build;
devstack update leaves such a binary alone).
The installer registers the dashboard hub to start at logon (on macOS a
LaunchAgent, for a Homebrew install too: not brew services) and a
devstack:// link the dashboard uses to start it. To skip both, set
DEVSTACK_NO_HUB_INTEGRATION=1 before the install command (Windows also
takes -NoHubIntegration when the script is run as a file); the choice is
kept, so devstack up does not register them later either. devstack hub install adds them, devstack hub uninstall removes them again.
Start
devstack init # a first devstack.toml from the repository (TODOs to work through)
devstack doctor # is this machine ready: Docker, Compose, tools, ports
devstack up # daemon, infrastructure, every autostart service
devstack ui # the dashboard
devstack skills install # skills for your coding agent, filled in from the stack file
A stack file names the Compose files of the infrastructure and the services devstack builds and runs:
version = 1
[stack]
name = "shop"
[compose.app]
project = "shop-dev"
files = ["deploy/compose.yml"]
services = ["db"]
[services.api]
build = "dotnet build src/Api -c Debug"
output = "src/Api/bin/Debug/net9.0"
run = "{artifact}/Api"
port = 5000
health = { http = "/health", timeout = "60s" }
watch = ["src/Api/**"]
env = { ConnectionStrings__Db = "Host={compose.app.db.5432}" }
[services.web]
kind = "devserver"
run = "npm run dev -- --port {port}"
port = 4200
The full reference: docs/stack-file.md; devstack schema prints the JSON Schema for your editor.
Every day
devstack status # every service; "not yet active" until your change runs
devstack journal # what changed at runtime, and through which client (dashboard, cli, mcp, tray)
devstack logs api --follow # --grep timeout across all, --since 2h for earlier instances
devstack restart api # rebuild now; the running one stays until the new one is healthy
devstack promote api # take over a candidate held back by a schema change
devstack rollback api # back to the build before the last replacement
devstack builds api # did my change ever go live, since when is it broken
devstack hot-reload api on # the toolchain's own watcher, for a debugging session
devstack run reset # a command of the stack file (a database reset, say)
devstack top --watch # CPU, memory, disk per service, container, group and lane
devstack env api # what a service gets, with origins
devstack logs --daemon # devstack's own decisions with their reasons
devstack down [--all] # containers go too (volumes stay); the daemon ends with the last tree; --all stops the lanes too
Lanes: every worktree, side by side
git worktree add ../shop-login -b login-fix # a worktree is a lane at once
cd ../shop-login && devstack up # its own builds, ports +100, its own containers
devstack lane start login-fix # the same, from the main tree
devstack lane list # every lane, where it runs, who has the main ports
devstack promote --lane login-fix # it answers on the main ports; --lane main gives them back
devstack lane stop login-fix
A lane runs from a copy of its worktree, so git can remove the worktree while it runs. It gets its own Compose project and data (seed jobs fill it), and no background workers unless it holds the main ports.
The pool: lanes on other machines
devstack join https://10.0.0.5:4260 --fingerprint <fp> # once; an admin approves
devstack up --on build-server # in a worktree: everything of the lane runs there
devstack up --on pool # the coordinator picks a node
devstack lane move login-fix --to local
devstack lane share login-fix --with team:qa
devstack open login-fix web # a lane someone shares with you (connects it)
devstack pool disconnect login-fix # its ports here are free again
One machine runs the coordinator, servers run a node agent; the dashboard's pool area admits members, shows capacity (GPUs included) over time, and sets what each node runs (who, which repositories, limits, network). Several pools per machine, a customer's beside the team's. Setup and everything else: docs/pool.md.
Agents
devstack mcp is an MCP server on stdio, a client of the same API as the CLI:
status, journal, up (with on for a node), down, start_service,
stop_service, restart_service, promote_service, rollback_service,
hot_reload, logs, builds, diagnostics, doctor, run_command,
usage, lanes, lane_start, lane_stop, lane_move, lane_share,
promote_lane, placement, hub_status, hub_start, instructions.
{ "mcpServers": { "devstack": { "command": "devstack", "args": ["mcp"] } } }
devstack skills install writes two skills into the repository
(.claude/skills/ by default; [agents] skills in the stack file names the
folders, .agents/skills for Codex and most other agents):
devstack (operating this stack, "is my change live", troubleshooting)
and devstack-config (editing the stack file). They are filled in from
the stack file; devstack status says when they are out of date. They are
local to the working tree, not committed: each folder ignores itself, and
the hub keeps them current for the agents switched on on its Skills page. A command with a confirmation text in the stack file runs only
when the agent passes confirm: true, which it may only do after you agreed.
Commands
Every command and flag: devstack --help, devstack <command> --help.
| Command | What it does |
|---|---|
devstack init, check [--compose], schema |
write a first stack file; validate it (with --compose every Compose group on its own too); print its JSON Schema |
devstack doctor |
what this machine lacks for this stack, with the fix; exits 1 when something fails |
devstack up [--on <node>|pool] [--workers] [--force] [--profile <name> [--run]] [--yes], down [--keep-infra] [--all], status |
start (in a worktree: its lane, here or on a node; with a profile of the stack file, for a purpose such as an acceptance run), stop with the containers (--all: the lanes too; --keep-infra, deprecated: leave them running); the daemon ends once no tree runs, show the stack |
devstack start|stop|restart <service|group>, promote, rollback, hot-reload |
one service or group; restart --rebuild builds again although an artifact for the same inputs exists |
devstack build <service> |
sync the main tree's snapshot and build an artifact, without starting it (in a lane: devstack restart <service> in its worktree) |
devstack builds <service>, logs, top, env <service> [--export dotenv] |
what ran, what it printed, what it uses, what it gets |
devstack run <command> |
a command of the stack file |
devstack tunnel [<service>] [--stop] [--all], --setup <file> --domain <d>, --pool <pool> |
a public link to a service: reachable from the internet by anyone with the address, without a login, until it is closed (ADR-0090); a Cloudflare Quick Tunnel, your own named tunnel, or a pool's. Needs cloudflared; see SECURITY.md |
devstack lane list|start|stop|reset|prune|move|share |
the lanes (worktrees), see above; reset throws a stopped lane's data away, prune removes what gone lanes left behind |
devstack ui [--lane <name>] [--browser <name|path>] [--link|--print] |
open and pair the dashboard, on this stack (in a worktree on its lane, or on --lane); the browser is --browser, else DEVSTACK_BROWSER, else BROWSER, else the dashboard's installed app when a Chromium-family browser has it (Linux, Windows; ADR-0061), else the system's default. --link prints the lasting links to the stack or lane instead (the page, and the devstack://open link that pairs first); --print the one-time pairing link. status, up and lane start end with the page as dashboard: <url> |
devstack hub [status|stop|start|install|uninstall] |
the hub that serves the dashboard: without an action in the foreground; stop leaves the stacks running; install registers it at logon and the devstack:// link (--no-autostart: the link only), uninstall removes both, and devstack up then leaves them alone |
devstack hub tray [on|off] |
the tray icon (ADR-0060): the state of every stack in the icon's colour, a menu to open the dashboard (in its installed app when there is one, as devstack ui), start or stop a stack and open its folder. The hub starts it (devstack tray); off ends it and keeps it off, on brings it back. Windows, Linux (a StatusNotifierItem: KDE, Xfce, GNOME with the AppIndicator extension) and macOS |
devstack skills install [--dir] [--force], skills check |
write the agent skills (the hub keeps them current on its own); check that git tracks none of them, a copy the hub leaves alone (exits 1 when it does, with the git rm -r --cached that takes it out) |
devstack mcp |
the MCP server (above) |
devstack join <url> --fingerprint <fp> [--name] [--pool] [--telemetry] [--no-wait] |
ask a pool to admit this machine (pool.md) |
devstack pool [shared|telemetry on|off], open <lane> [<service>] |
the pool as this machine sees it; lanes others share with you |
devstack node init|serve|fingerprint|agent |
run a pool's coordinator or a node agent (pool.md) |
devstack node members [approve|remove|role|tags|node|policy] |
administer a pool on its coordinator; node <id> on|off grants a member to run the pool's lanes |
devstack update [--check] [--version <v>], self uninstall [--with-coordinator] |
install a newer release (or only say whether one exists) and move the hub onto it; remove devstack for this user (its pool memberships go too; a coordinator kept in the per-user state only with --with-coordinator) |
Why
Most local orchestrators run a service straight out of the working tree's build output and restart it whenever a file changes. With one developer that is tolerable. With an IDE, several terminals and several AI agents editing the same tree, it gives a stack that is half-built, file-locked, restarted three times a second, and impossible to ask "is my change running?". docs/problem-statement.md describes the failure modes devstack removes.
Shape
devstack CLI ──┐ ┌── infrastructure: docker compose, health-gated
MCP (agents) ──┼──► devstackd ──────┼── builds: snapshot → build → artifact <build-id>
dashboard ─────┘ one per repo, └── services: main slot (fixed ports) + candidates,
│ loopback API lanes (worktrees) on their own ports
└── hub (every stack, the pool) ──► coordinator ──► node agents (lanes of others)
Environment
| Variable | Effect |
|---|---|
DEVSTACK_BROWSER |
the browser devstack ui pairs (chrome, edge, firefox, brave, a path), and that a click on a notification or the tray opens; set, it wins over the dashboard's installed app |
DEVSTACK_HUB_PORT |
the hub's port (default 4250) |
DEVSTACK_NO_HUB |
devstack up does not start the dashboard hub |
DEVSTACK_NO_HUB_INTEGRATION |
the installer registers no hub at logon and no devstack:// link (and records that); devstack up registers neither while it is set |
DEVSTACK_NO_TRAY |
the hub does not start the tray icon (devstack hub tray off keeps it off for good) |
DEVSTACK_NO_NOTIFY |
the hub shows no desktop notifications (they are set per kind with devstack hub notify or the bell in the dashboard) |
DEVSTACK_NO_UPDATE_CHECK |
no check for a newer release |
DEVSTACK_OTLP_ENDPOINT |
devstack's own diagnostic log also to this OTLP/HTTP collector |
DEVSTACK_HUB_PING_ORIGINS |
more origins (comma-separated) that may ask whether a hub runs, for a copy of the landing page |
DEVSTACK_POOL_DIR, DEVSTACK_COORDINATOR_DIR, DEVSTACK_TELEMETRY_* |
the pool: see docs/pool.md |
Documentation
| docs/stack-file.md | the stack file, every key |
| docs/pool.md | the pool: coordinator, nodes, lanes on other machines, sharing, policies |
| docs/dashboard-api.md | the hub's and the daemon's HTTP API |
| docs/plan.md | architecture, phases, what is open |
| docs/adr/ | the decision log |
| CHANGELOG.md | what changed in each release |
| docs/releasing.md | how a release is made: the checklist |
| SECURITY.md | what devstack exposes, what leaves the machine, the pool's trust model, reporting a vulnerability |
Contributing
Documentation, code and commits in English. Architectural decisions are recorded as ADRs before they are built (AGENTS.md). The project is generic: downstream stacks keep their stack file, Compose files and secrets in their own repositories.
License
MIT or Apache-2.0, at your option (LICENSE-MIT, LICENSE-APACHE). Security reports: SECURITY.md.
Preguntas sobre devstack MCP
¿Qué puede hacer el MCP server de devstack?
El MCP server permite a los agentes realizar operaciones como comprobaciones de estado, iniciar/detener servicios, gestionar lanes, ver logs y ejecutar diagnósticos.
¿Cómo gestiona devstack los cambios de código?
Los cambios se compilan como candidatos junto a la instancia en ejecución y solo se implementan una vez que se confirma que son estables, evitando builds rotos.
¿Puedo ejecutar mi stack de desarrollo en un servidor remoto?
Sí, puedes ejecutar el lane de un worktree en una máquina dentro del pool de un equipo con sus puertos redirigidos a tu máquina local.