KARIMO plugin
A Claude Code plugin that takes a feature from PRD to merge: sub-agents work in waves on separate branches, each PR gets a review
- 293 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- Agents write code, create branches and worktrees in your repository and open pull requests through GitHub CLI
- Running several agents in parallel burns through your plan's limits and tokens quickly
- The plugin installs lifecycle hooks that can connect to Slack and Jira
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/karimoDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Agents write code, create branches and worktrees in your repository and open pull requests through GitHub CLI
- Running several agents in parallel burns through your plan's limits and tokens quickly
- The plugin installs lifecycle hooks that can connect to Slack and Jira
This is third-party code. Review the repository files before installing.
What it does
KARIMO takes a feature from an idea to a merge into the main branch. /karimo:research creates a PRD folder and researches the codebase, and /karimo:plan runs an interview and records requirements. /karimo:run splits them into task briefs, checks the briefs against the code and starts execution, and /karimo:merge opens the final pull request. Tasks run in waves: inside a wave agents work in parallel, each in its own git worktree, and the waves follow task dependencies. The plugin verifies that an agent sits on the right branch and picks a model by task complexity, escalating to a stronger one on failure. It catches loops by the meaning of repeated actions. After a crash it reconciles git state with the plan and continues. Each pull request can be reviewed by hand or by automated review. /karimo:dashboard shows progress, /karimo:feedback saves lessons for later runs and /karimo:doctor diagnoses the installation.
Who it is for. For developers and designers who ship large features in Claude Code from a PRD and want parallel agents with a checkpoint at every step.
Good fit when
- A feature is too big for one session and needs tasks with dependencies
- You want several agents in parallel without mixing their changes: each gets its own worktree and branch
- Every pull request should pass Claude Code Review or Greptile before merging
- You prefer to capture requirements in an interview first and hand execution to agents afterwards
Not a fit when
- The change is a fix in one or two files: the interview and briefs cost more than the task
- You use an agent other than Claude Code: the plugin is built on its sub-agents, hooks and worktrees
- Agents must not write to the repository or create branches
Example request
/karimo:research "CSV report export", then /karimo:plan --prd export-csvLimitations
Requires Claude Code, GitHub CLI with gh auth login completed and Git 2.5 or newer. Automated review in the second phase needs Greptile or Claude Code Review; without them checks stay manual. The author puts Greptile at 30 dollars a month. The author advises caution when running several feature branches with worktrees at once: on the main branch it is best to only research and plan. The plugin installs from the repository's own marketplace for now, while its application to Anthropic's official marketplace is under review.
How to disable. Run /plugin and disable or remove karimo. If you used the legacy install script, delete the .claude/plugins/karimo folder in the project.
FAQ
How do I install KARIMO in Claude Code?
Run /plugin marketplace add opensesh/KARIMO, then /plugin install karimo@karimo and /reload-plugins. The older .karimo/install.sh script still works but the author no longer recommends it.
Can KARIMO run without automated review?
Yes. Review belongs to the second adoption phase. PRD interviews, agent execution and pull requests work right after installation.
Do I need a separate feature branch?
No. /karimo:configure sets the mode: tasks branch from a feature branch or directly from the main branch.
How is KARIMO different from Claude Code's built-in sub-agents?
Claude Code provides worktrees, task spawning and hooks. KARIMO adds branch verification for each agent, wave ordering by dependencies, model choice by complexity, git state reconciliation after a crash and loop detection.
Which commands does KARIMO have beyond the main cycle?
/karimo:configure sets up branching and /karimo:greptile-review runs a Greptile review. There are also /karimo:help and /karimo:update.
Discussion
No questions or reviews yet.
Related
PluginsEleven skills take an agent from an idea to verified code through a PRD and RFCs kept as files in the repository
A Claude Code plugin: 39 commands for PRs, review, commits and specs, nine expert roles and hooks, in eight language versions
A Claude Code plugin that hands a task to a local no_human and reviews a branch with a reviewer that never saw the coder's session
A universal workflow plugin for Claude Code: 7 agents, 26 commands, 14 skills and 14 hooks for any project
A Grid Dynamics plugin that loads a team's shared engineering standards, architecture and workflows into a coding agent session
Goal Workflow
goal-workflow
19 skills for Claude Code: PRD, spec, GitHub Issues, batch implementation with review and pull request merging, plus refactoring and diagrams
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| KARIMO (this entry) | no | 293 | |
| oh-my-claudecode | no | 40,377 | team orchestration and autopilot without PRD interviews and dependency waves |
| The Agentic Startup | no | 575 | plugins for spec-driven development, while KARIMO focuses on parallel waves and review of every pull request |
README in short
The process has six steps, from research to pull request inspection. The author suggests adopting the plugin in three phases: the basic run, automated review and a console dashboard with velocity metrics.