SpaceApp Super Agent Workspace

Self-hosted Docker workspace where up to 16 AI coding CLIs run side by side in persistent rooms, and Nous orchestrates them toward your goal.

Source repo
oll4com/spaceapp
Stars
★ 20
Last updated
1d ago
License
MIT
Primary language
TypeScript

At a glance

How it runs
CLISelf-hosted serviceWeb app
Works with
Universal · cross-platformCodex · Claude Code · OpenAI API · Claude APIClaude.ai (Partial support)
Cost
Free software; you pay for model usage
Setup effort
Medium · a few setup steps
You'll need
DockerNode.js >=26.13 <27 (development)PostgreSQL with pgvectorShell / CLINetwork accessLocal filesystemMCP Server
Typical use
Developers who want to compare OpenCode, Claude Code, Codex and Gemini on the same task side by side in one room.
Not a fit if
  • Solo users who want a fully managed cloud service instead of running Docker and Postgres themselves
  • Teams needing strong multi-tenant isolation between mutually untrusted users
  • Users unwilling to sign in with their own Claude Code or Codex provider accounts
Website
spaceapp.dev ↗ set on GitHub
Maintenance
Commits in the last 30 dayslast commit 2026-11-20latest release v1.2.2 (2026-11-14)

What does this agent do, and when should you use it?

SpaceApp is a Docker Compose web app you run on your own machine that replaces a wall of terminal windows with rooms of panes, where each pane is an AI coding CLI, a copilot, a browser or a tool. It advertises up to 16 concurrent coding assistants — OpenCode, OpenAI Codex, Claude Code, Google Gemini, Qwen, Kimi, Grok, DeepSeek and others — in rooms that stay warm so sessions keep running when you switch away. Its own agent, Nous, reads each pane's progress, sends next steps and works on /goal objectives, asking you before releases, deletes, third-party or money actions and changes of direction. Every pane shares Space Memory, a credential Vault, Space Gateway and skills/MCP servers. Installation is one npx command: the launcher checks CPU, RAM and disk, validates image SBOM and provenance attestations, pulls by pinned digest and starts the small, medium or large profile. The stack is a TypeScript monorepo (React web, Fastify API, Temporal worker) under MIT.

Running npx --yes run-spaceapp@latest install checks hardware, installs or starts Docker, verifies SBOM and provenance attestations of SpaceApp's images, pulls them by the digest the npm package pins and starts the stack, then prints a 15-minute setup token and opens http://127.0.0.1:4911. Inside, spaceapp-core serves the web app and API, spaceapp-cli hosts the AI CLIs, postgres holds data and pgvector memory; the medium profile adds temporal workflows and large adds spaceapp-browser (managed Chrome). Containers see no host files until you run workspace add <path> (--read-only for reference repos); provider install <name> adds CLIs such as codex, claude, gemini or qwen, and credentials set <provider> stores API keys via masked input. Each pane is a real signed-in CLI session; Nous reads their output turn by turn, and once Claude Code is connected you can enable Work on my behalf in the Nous dock so it executes goal turns itself (each turn is a headless Claude Code run on your own login). Operational commands include up/down/status/logs, update/rollback, backup/restore, doctor --fix and owner reset-password.

  1. Developers who want to compare OpenCode, Claude Code, Codex and Gemini on the same task side by side in one room.
  2. Teams that must keep source and prompts on their own hardware, deploying via Docker locally or on an internal server and connecting from a browser or phone.
  3. Users who want to hand a longer objective to Nous via /goal and only approve releases, deletes or spending.
  4. Solo developers and small teams that also want Mail, Job, Marketing, Freelance, Security and Monitoring copilots enabled as optional modules.
  5. Project groups that want multiple agent sessions reading and writing one shared decision record through Space Memory and MCP servers.

Key setup facts

Pulled from the install and usage notes below. The project README has the current values.

Install / run
  • npx --yes run-spaceapp@latest install
  • npx --yes run-spaceapp@latest install --profile small
  • npx --yes run-spaceapp@latest status
  • npx --yes run-spaceapp@latest workspace add ~/projects/demo
Config files
config.json
Install notes cover
Windows · macOS · Linux · Docker

How do you install or deploy this agent?

One command on Linux, macOS or Windows 11; Docker is installed automatically if missing:

npx --yes run-spaceapp@latest install

On Windows use Command Prompt; if PowerShell blocks scripts, run:

npx.cmd --yes run-spaceapp@latest install

You can also pin a resource profile explicitly:

npx --yes run-spaceapp@latest install --profile small

Minimums: small needs 4 CPUs, 7 GiB RAM and 6.5 GiB disk; medium needs 11 GiB RAM and 8 GiB disk; large needs 8 CPUs, 15 GiB RAM and 11 GiB disk.

How do you use this agent?

The launcher prints a setup token and opens http://127.0.0.1:4911; paste the token, create your owner account, and the Getting Started room opens with OpenCode. Everyday commands:

npx --yes run-spaceapp@latest status
npx --yes run-spaceapp@latest workspace add ~/projects/demo
npx --yes run-spaceapp@latest provider install claude
npx --yes run-spaceapp@latest credentials set anthropic
npx --yes run-spaceapp@latest doctor --fix

To enable optional modules, set optionalModules in the launcher's config.json, then run:

npx --yes run-spaceapp@latest repair

In Nous, type /goal <what you want>; after connecting Claude Code, switch on Work on my behalf in the Nous dock to let it act.

What are this agent's strengths and limitations?

Pros
  • A single npx command installs everything: hardware checks, Docker bootstrap, and digest-pinned pulls verified against SBOM and provenance attestations.
  • Runs up to 16 coding CLIs side by side in persistent rooms, each using your own provider account rather than a shared vendor credential.
  • Nous acts as one orchestrator that reads every pane and drives /goal objectives, with hard stops before releases, deletes, third parties and money.
  • Shared infrastructure across panes: masked Vault credentials, pgvector-backed Space Memory, Space Gateway and MCP servers.
  • Auto-selects small/medium/large profiles by hardware and ships backup/restore, update/rollback and doctor --fix for day-two operations.
Limitations
  • Containers are explicitly not a sandbox against hostile workloads, and rooms are not a security boundary between mutually untrusted users.
  • Heavy footprint: Docker plus PostgreSQL/pgvector, and medium/large profiles need 11–15 GiB RAM plus Temporal and managed Chrome.
  • Copilots and Studios are optional modules that require editing config.json and running repair; they are not on by default.
  • Self-hosted Nous depends on your own Claude Code login and headless Claude Code runs, so capabilities track that CLI's versions and credentials.
  • License history changed from Apache-2.0 to BSL 1.1 and only became MIT at 1.1.7, so older releases carry different terms.
  • No AGENTS.md or repository guidance file is provided for agent-assisted customization.

How does this agent compare with similar options?

The README names no specific competitor; it positions itself against juggling a wall of terminal windows and differentiates by co-hosting existing CLIs (OpenCode, Claude Code, Codex, Gemini, Qwen and others) rather than replacing any single one.

Key facts side by side with the most closely related agents.

Agent Source review Form / cost Stars Updated Language Full support on
SpaceApp Super Agent Workspace This agent 55 · Major gaps CLIFree + model costs ★ 20 1d ago TypeScript Codex · Claude Code · OpenAI API · Claude API
OtoDock 70 · Some gaps Self-hosted serviceFreemium ★ 202 today Python Codex · Claude Code
Codeg Multi-Agent Coding Workspace 65 · Some gaps Desktop appFree + model costs ★ 4.5k 1d ago Rust Codex · Claude Code · OpenAI API · Claude API
Synapse AI — Multi-Agent Orchestration Platform 58 · Major gaps CLIFree + model costs ★ 343 3d ago Python OpenAI API · Claude API

How does FollowAgents rate this agent?

FollowAgents source review · FARS-2.1
Major gaps
55/ 100 5-point scale 2.8 / 5
This review covers the source as of Nov 12, 2026. The project has shipped 2 stable release(s) since, which have not been re-reviewed. The score and verdict describe the reviewed version only. Recent releases ↓
Trust 18/29
Reliability 6/14
Adaptability 10/18
Convention 10/18
Effectiveness 7/13
Verifiability 4/8
Why each dimension lost points
Trust18 / 29 · 3.1/5

The security model documents loopback binding, one-time setup token, no Docker socket in containers, no-new-privileges, keys mounted only to the CLI service, telemetry off by default, and explicit host-root risk plus a revert path — solid least-privilege and sensitive-data handling. However host-root mode mounts host / read/write with no per-command approval, a high-risk optional capability; external effects (installing Docker, adding user to docker group, UAC/restart, Sentry reporting) are documented but heavy by default, so external_effects scores 1. Dependency security has pinned overrides and digest/SBOM/attestation checks but no full vulnerability-scan evidence, scoring 2. Rollback has update/rollback/backup/restore, scoring 2. Source attribution has LICENSE, THIRD_PARTY_LICENSES and UPSTREAM.md references, scoring 2.

Reliability6 / 14 · 2.1/5

Internal consistency is decent (security model, install docs and CI workflow corroborate each other), but KNOWN_LIMITATIONS admits public CI has long been non-green with flaky tests and an open P0, so reliability evidence is incomplete. Dependency availability is weakened by engines requiring Node >=26.13 <27 while getting-started says Node 20.11+, a documented conflict, scoring 1. Failure messages show explicit error codes (ADMIN_REQUIRED, DIAGNOSTICS_BUSY) and refusal paths, but evidence is sample-level only, scoring 1.

Adaptability10 / 18 · 2.8/5

Targets a single-owner self-hosted scenario with small/medium/large profiles and multi-platform install notes; capability boundaries are explicit (not multi-tenant, not a sandbox, host-root dangerous). Trigger precision is weak: sampled SKILL.md files are internal ops/benchmark workflows tied to public-host and space_ops, limiting fit for ordinary users, scoring 1. Environment fit across Linux/macOS/Windows scores 2.

Convention10 / 18 · 2.8/5

Install notes are very thorough (getting-started, installer-release-runbook, command table), scoring 3. Information architecture is clear, scoring 2. Naming stability is weak: skill renames (job-applications→space-job-applications) and multiple license changes (Apache→BSL→MIT), scoring 1. Examples and FAQ are sparse, mostly command snippets, scoring 1. Known limitations are detailed but list many open items, scoring 2. License is MIT and complete, scoring 2. Versioning/changelog is scattered with no standalone CHANGELOG, scoring 1. Maintenance responsibility is limited: publisher is unverified and SECURITY.md gives a reporting flow but modest commitments, scoring 1.

Effectiveness7 / 13 · 2.7/5

Output usability is high (rooms/panes, Nous, command set), but cost is high (Docker stack, 7-15 GiB RAM, multiple containers); marginal value depends on already needing many CLIs, so cost_benefit scores 1. Output usability and marginal value score 2 each.

Verifiability4 / 8 · 2.5/5

Many claims carry claims annotations and audit docs (claims-audit, runtime-compatibility), giving decent traceability, scoring 2. Cross-source corroboration is limited: CI status is self-reported as non-green and some claims rely on external links, scoring 1. Fact/inference separation is explicit in KNOWN_LIMITATIONS (fixed vs open vs not-a-defect), scoring 2.

Risks and how to mitigate them
  • host-root mode mounts host / read/write with no per-command approval; use only on a disposable, trusted Linux machine.
  • The installer auto-installs Docker, adds the user to the docker group (root-equivalent), and may trigger UAC or a restart; run in a controlled environment.
  • engines requires Node >=26.13 <27 while getting-started says Node 20.11+, a version conflict that may break installs.
  • KNOWN_LIMITATIONS admits public CI has long been non-green with flaky tests and an open P0; reliability evidence is incomplete.
  • Publisher identity is unverified by FollowAgents; maintenance commitments and update path need independent confirmation.
  • Skill naming and license history changed multiple times; confirm which terms apply to the current version before integrating.
Evidence confidence: Low Reviewed Nov 12, 2026 Reviewed revision e287cde40b0b New commits since this review; the score may not cover them
See the full review method →
Maintain this project? Add the FARS badge to your README

The badge shows the current review score and verdict and links back to this page. It refreshes about once an hour after a re-review.

FARS

[![FARS review](https://followagents.com/badge/spaceapp.svg)](https://followagents.com/en/agents/spaceapp)
<a href="https://followagents.com/en/agents/spaceapp"><img src="https://followagents.com/badge/spaceapp.svg" alt="FARS review"></a>

Recent releases

Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.

  1. v1.2.2 SpaceApp setup 1.2.2
    Release notes excerpt
    SpaceApp setup 1.2.2 is a patch release: security updates, a lighter browser host on small machines, and launcher fixes. Install: npx --yes run-spaceapp@latest · Update: npx --yes run-spaceapp@latest update Source: commit f5340c1ee1119a9718c004301746e8af7f578a15. npm [email protected]; images ghcr.io/oll4com/spaceapp-core:1.2.2 (sha256:ec583c4488f72a924c7c1913f090912b8c4e43d3633d18d0f7674b17752229c5), ghcr.io/oll4com/spaceapp-cli:1.2.2 (sha256:73aa7ed6ca70671ec292aa8ed380683f876f54290c1945e0ee612789ee037880), ghcr.io/oll4com/spaceapp-browser:1.2.2 …
    View on GitHub ↗
  2. v1.2.1 SpaceApp setup 1.2.1
    Release notes excerpt
    SpaceApp setup 1.2.1 is a patch release: security updates and installer fixes. Install: npx --yes run-spaceapp@latest · Update: npx --yes run-spaceapp@latest update Source: commit 10072d5e8e2eef787ff617a36c7dfe6f45ad3d2d. npm [email protected]; images ghcr.io/oll4com/spaceapp-core:1.2.1 (sha256:7b23436a477e05a5c7dc10e07d93f8d7b21be5eb486d81ba6ff038042106762e), ghcr.io/oll4com/spaceapp-cli:1.2.1 (sha256:243694f41585c4e6ba8a4471de4dde94ba494c11aa440431adb5329873caa1b2), ghcr.io/oll4com/spaceapp-browser:1.2.1 …
    View on GitHub ↗
  3. v1.2.0 SpaceApp setup 1.2.0
    Release notes excerpt
    SpaceApp setup 1.2.0 brings Node.js 26 and TypeScript 7, Nous on self-hosted installs, and the fixes from the 1.1.9 in-app reviews. Install: npx --yes run-spaceapp@latest · Update: npx --yes run-spaceapp@latest update Source: commit 748605c2148db8583076faa69e8da044b42e9fb6. npm [email protected]; images ghcr.io/oll4com/spaceapp-core:1.2.0 (sha256:c99c153a0b36f668e828b9410a37eb0555ae5adb68148ac50c8abdb7a3244876), ghcr.io/oll4com/spaceapp-cli:1.2.0 (sha256:30991e575830d0deb0f43aca6fe341f007e9f657f7cf7a630720c930847764de), ghcr.io/oll4com/spaceapp-browser:1.2.0 …
    View on GitHub ↗

FAQ

What does SpaceApp itself cost?
The software is MIT-licensed and self-hosted. The real cost is the AI coding CLIs and model usage you run, which go through your own provider accounts or API keys; the README names no paid hosted plan.
Can agents read my host files or see my credentials?
Containers see no host files until you register a folder with workspace add <path> (--read-only for reference repos). The Vault fills logins without revealing them to agents, but the README states containers are not a sandbox against hostile workloads and rooms are not a security boundary.
How autonomous is Nous?
It is off until you enable Work on my behalf in the Nous dock, and it asks before releases, deletes, third-party or money actions and direction changes. On a self-hosted install each turn is a headless Claude Code run on your own Claude Code login.
What machine do I need?
small needs 4 CPUs, 7 GiB RAM and 6.5 GiB disk; medium adds Temporal and needs 11 GiB RAM; large adds managed Chrome and needs 8 CPUs, 15 GiB RAM and 11 GiB disk. The launcher picks the largest profile that fits, or you pass --profile <name>.
Can I roll back a bad update?
update replaces the images and rollback returns to the previous ones, but database changes persist, so schema-level recovery requires the restore command with a checksummed backup archive of database, settings and memory.
Is SpaceApp Super Agent Workspace open source?
Yes. The source code is on GitHub (oll4com/spaceapp) under the MIT licence.
Is SpaceApp Super Agent Workspace safe to use?
FollowAgents' FARS review (Nov 12, 2026) scored SpaceApp Super Agent Workspace 18/29 on trust and safety. None of the eight safety controls the review checks for was missing outright. The repository has changed since that review.
Is SpaceApp Super Agent Workspace still maintained?
Yes. The latest commit on GitHub is from 2026-11-20, within the last 30 days. The latest release is v1.2.2 (2026-11-14).
Who makes SpaceApp Super Agent Workspace?
SpaceApp Super Agent Workspace is published on GitHub by the account oll4com. FollowAgents has not verified which person or company is behind that account.
What is SpaceApp Super Agent Workspace's official website?
The project's GitHub repository lists https://spaceapp.dev as its website. That link is set by the repository owner; the source code itself is on GitHub (oll4com/spaceapp).
Was this page helpful?
View on GitHub ↗ Install ↓

Host ecosystems

Compare agents like this one

The same FARS review applied across the shortlist this agent qualifies for.

Related concepts