---
title: "Spacedock · nothing ships without a decision"
description: "Spacedock is a multi-agent orchestrator where nothing ships without a decision. It lives within your existing harness: Claude Code or Codex."
doc_version: "0.0.1"
last_updated: 2026-06-11T23:48:24.183Z
canonical: https://spacedock.md/
---

# Spacedock

> Spacedock is a multi-agent orchestrator where nothing ships without a decision. It lives within your existing harness: Claude Code or Codex.  Spacedock pools the decisions your agents need from you into one evidenced queue, so you make the few calls that matter and the rest keeps running.

## What it is

- The unit is the decision: evidence in, a resolution out (approve / send back / escalate), reason recorded.
- A separate agent reviews the work, so no agent rubber-stamps itself.
- Good is discovered, not declared: the bar starts rough and sharpens as you reject.
- Brownfield-first: on install it scans ~a month of your sessions and surfaces the workflow you already run.

## Who it's for

- Heavy Claude Code and Codex users with brownfield projects, accountable for several agentic work streams.

## What it's not

- Not pure automation
- Not generic review
- Not an eval suite
- Not one agent grading itself

## Harnesses

- Runs inside: Claude Code, Codex

## Sitemap

- [Landing page](https://spacedock.md/): what Spacedock is, early access form
- [Agent guide](https://spacedock.md/agents.md): how an agent should use this site
- [Markdown sitemap](https://spacedock.md/sitemap.md)
- [llms.txt router](https://spacedock.md/llms.txt)
- [Docs](https://spacedock.md/docs/): full markdown index at https://spacedock.md/docs/llms.txt
- [Source](https://github.com/spacedock-dev/spacedock)