multi-repo-swarm — Claude Code subagent: repository orchestration
multi-repo-swarm — a subagent for Claude Code that orchestrates multiple agents across your organization's repositories. Runs synchronized tasks, manages dependencies, and links pull requests between projects.
no license ★ 850 updated 2026-10-26
.claude/agents/multi-repo-swarm.mdWhat multi-repo-swarm does
Sets up a swarm with hierarchical topology where each repository gets a role—UI, API, library—and its own set of specialized agents. Synchronizes operations through GitHub CLI: discovers repos by language and dependencies, applies the same changes everywhere (dependency updates, refactoring, security patches), and creates linked pull requests.
Uses swarm_init, agent_spawn, task_orchestrate tools for lifecycle management, plus github_repo_analyze, github_pr_manage, github_sync_coord to work with repositories. Requires GitHub CLI authentication—gh auth status is verified on startup. Configure via .swarm/multi-repo.yml that describes repos, their roles, interdependencies, and sync strategy.
What it helps with
- update a single dependency across all microservices simultaneously
- run large-scale refactoring coordinated between frontend and backend projects
- deploy a security patch to all affected repositories and verify test results
- track and sync pull requests linked to a single feature across different projects
Example prompts for the agent
- Run multi-repo-swarm to upgrade TypeScript to 5.0.0 in all repositories with TypeScript in their stack
- Coordinate refactoring from OldAPI to NewAPI: analyze impact across all projects and generate a migration guide
- Scan all repositories for vulnerabilities, apply patches, and create a compliance report
How to install multi-repo-swarm
Manually: put the files in place
.claude/agents/multi-repo-swarm.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention
Manually: put the files in place
.codex/agents/multi-repo-swarm.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/multi-repo-swarm.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about multi-repo-swarm
How do I install multi-repo-swarm in Claude Code?
Add the subagent definition to .claude/agents/multi-repo-swarm.md at your project root. Claude Code will read it on startup and load the orchestration tools.
What does multi-repo-swarm do?
It manages multiple agents across different repositories in your organization at once. Use it to make synchronized changes (updates, patches, refactoring) to all projects simultaneously instead of repeating the same operations manually.
What do I need before I can run it?
GitHub CLI must be authenticated. Run gh auth login if you haven't already. The subagent checks this on initialization and will not start without access.