Ralph Swarm
- 2 repo stars
- Updated February 22, 2026
- Divkix/ralph-swarm
ralph-swarm is a Claude Code plugin that turns a development goal into a multi-section spec and then executes tasks with Agent Teams sequentially or in parallel.
About
ralph-swarm is a Claude Code plugin marketplace entry for spec-driven development with Agent Teams. It first generates a structured spec across research, requirements, design, and tasks, then executes those tasks with specialized agents in parallel or sequence. Use it for medium or large features where you want review gates by default and optional yolo mode when a written spec should drive autonomous implementation.
- Single command pipeline: Goal through Research, Requirements, Design, Tasks, then Execution
- Optional --full autopilot or review pauses between planning sub-phases
- Executes tasks sequentially or in parallel with Agent Teams and worktrees
- Claude Code plugin marketplace entry ralph-swarm@ralph-swarm
Ralph Swarm by the numbers
- Data as of Jul 7, 2026 (Skillselion catalog sync)
/plugin marketplace add Divkix/ralph-swarmAdd your badge
Show developers this marketplace is listed on Skillselion. Paste this into your README.
| repo stars | ★ 2 |
|---|---|
| Last updated | February 22, 2026 |
| Repository | Divkix/ralph-swarm ↗ |
How do you get a reviewed spec and coordinated multi-agent execution from one Claude Code command?
Run ralph-swarm to turn a goal into research, requirements, design, and tasks then execute via Agent Teams in Claude Code.
Who is it for?
Engineers tackling sizable features in Claude Code who want documented specs plus Agent Teams execution with review checkpoints.
Skip if: Tiny one-file edits where formal spec phases and team orchestration add more overhead than value.
What you get
A completed spec artifact set and implemented tasks after the SWARM COMPLETE execution phase.
Plugins in this marketplace
1 plugin - install individually after you add the marketplace.
Recommended Marketplaces
FAQ
What are the two phases?
README defines Planning (research, requirements, design, tasks without code) then Execution where agents write code.
How does review work by default?
Each planning sub-phase pauses for review unless you pass --full to run all planning phases at once.
Who is it aimed at?
README lists teams and developers wanting repeatable planning before coding on medium or large parallelizable tasks.