Multi-agent orchestration system for Claude Code. HOUSTON coordinates Voyages, Missions, and Objectives while fresh Pods execute work autonomously.
npx claudepluginhub thebrownproject/space-agents --plugin space-agentsPropose architectural approaches with trade-offs during brainstorming
Explore codebase for relevant patterns, constraints, and context during brainstorming
Identify risks, failure modes, and effort estimates during brainstorming
Trace code paths, gather evidence, and investigate issues during debugging sessions
Writes code for tasks using Pathfinder findings and TDD approach
Reviews implementation for requirements and code quality (two-pass)
Explores codebase and provides implementation context for tasks
Create detailed TDD task breakdown for tasks
Analyze dependencies and sequence tasks during planning
Break a feature into tasks during planning
Simplifies and refines code for clarity, consistency, and maintainability while preserving all functionality. Focuses on recently modified code unless instructed otherwise.
Review code for performance issues and optimization opportunities
Review code for quality, readability, and maintainability issues
Review code for security vulnerabilities and risks
Check mission status and progress via fresh subagent. Queries Beads for features, tasks, bugs, and recent work activity. Keeps HOUSTON context lean.
Explore ideas through conversation. HOUSTON asks questions, has opinions, and suggests background agents when investigation would help.
Interactive debugging through conversation. HOUSTON guides through 4 phases, asks questions, and suggests agents for investigation.
Structure work into plans and Beads. Three modes: plan from brainstorm, plan from scratch, or create Beads from existing plan.
Interactive code review through conversation. HOUSTON guides review, spawns specialized agents, and helps create Beads for issues found.
Produce structured spec.md from brainstorm conversations. Creates a consistent format that feeds cleanly into the planning phase.
Select mode, delegate to analysis skill.
Use to install Space-Agents in a project. Creates .space-agents/ directory and initializes Beads for issue tracking. Run once per project.
End session. Reconciles exploration folders, appends context to CAPCOM, syncs Beads, displays session stats.
Start a Space-Agents session. Displays welcome screen with project status.
Run test and lint validation gate. Use after implementation to verify code quality.
HOUSTON spawns Pathfinder, Builder, Inspector per task. Best for medium features (4-10 tasks).
Execute a single task with Pathfinder/Builder/Inspector crew. Self-fetches work from Beads.
Launch ralph.sh in background. Lightweight mode (default) or full Pod crew (--pod). Best for large features (10+ tasks).
HOUSTON executes feature tasks directly. No agents. Best for small features (1-3 tasks).
Select feature, select mode, delegate to execution skill.
███████╗██████╗ █████╗ ██████╗███████╗
██╔════╝██╔══██╗██╔══██╗██╔════╝██╔════╝
███████╗██████╔╝███████║██║ █████╗
╚════██║██╔═══╝ ██╔══██║██║ ██╔══╝
███████║██║ ██║ ██║╚██████╗███████╗
╚══════╝╚═╝ ╚═╝ ╚═╝ ╚═════╝╚══════╝
█████╗ ██████╗ ███████╗███╗ ██╗████████╗███████╗
██╔══██╗██╔════╝ ██╔════╝████╗ ██║╚══██╔══╝██╔════╝
███████║██║ ███╗█████╗ ██╔██╗ ██║ ██║ ███████╗
██╔══██║██║ ██║██╔══╝ ██║╚██╗██║ ██║ ╚════██║
██║ ██║╚██████╔╝███████╗██║ ╚████║ ██║ ███████║
╚═╝ ╚═╝ ╚═════╝ ╚══════╝╚═╝ ╚═══╝ ╚═╝ ╚══════╝
Agent orchestration plugin for Claude Code, powered by Beads and the Ralph loop.
npm install -g beads-cli)npm install -g mprocs) for watching Ralph loop live1. Add the plugin (in Claude Code):
/plugin marketplace add thebrownproject/space-agents
/plugin install space-agents@thebrownproject-space-agents
2. Initialize in your project:
/install
This creates the .space-agents/ directory structure and initializes Beads for issue tracking. Run once per project.
/launch → /exploration → /mission → /land
| Command | Purpose |
|---|---|
/launch | Start session, see project status |
/exploration | Brainstorm, plan, review, or debug |
/mission | Execute work from Beads |
/land | End session, save context to CAPCOM |
| Mode | Output |
|---|---|
brainstorm | Ideas and analysis in exploration/ideas/ |
plan | Implementation plans in exploration/planned/ |
review | Code review findings as Beads |
debug | Investigation results as Beads |
| Mode | Best for |
|---|---|
solo | Small work (1-3 tasks). HOUSTON executes directly |
orchestrated | Medium work (4-10 tasks). Agent per task |
ralph | Large work (10+ tasks). Background automation |
In orchestrated and ralph modes, each task is executed by a 3-agent crew:
Pathfinder → Builder → Inspector
| Agent | Role |
|---|---|
| Pathfinder | Explores codebase, gathers context, identifies patterns and risks |
| Builder | Implements the task using TDD approach |
| Inspector | Two-pass review: requirements verification, then quality/patterns/security |
The Ralph loop spawns these as Pods with fresh context each iteration, looping until all tasks complete or a blocker halts execution. When issues are found, bugs are inserted into Beads as blockers; Ralph resolves these before continuing to the next task.
Complete collection of battle-tested Claude Code configs from an Anthropic hackathon winner - agents, skills, hooks, rules, and legacy command shims evolved over 10+ months of intensive daily use
Uses power tools
Uses Bash, Write, or Edit tools
Complete collection of battle-tested Claude Code configs agents, skills, hooks, rules, and legacy command shims evolved over 10+ months of intensive daily use
Team-oriented workflow plugin with role agents, 27 specialist agents, ECC-inspired commands, layered rules, and hooks skeleton.
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
AI-powered development tools for code review, research, design, and workflow automation.
AI-supervised issue tracker for coding workflows. Manage tasks, discover work, and maintain context with simple CLI commands.