Claude Lane Stack plugin
A one-person coding factory: Claude Code manages the project while Codex, Qwen, Grok and Kimi write code as separate processes.
- 125 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The manager merges results into `main` by itself, so a writer mistake can reach the main branch.
- Writers run as background processes and execute code in the project workspace.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/claude-lane-stackDetects 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.
- The manager merges results into `main` by itself, so a writer mistake can reach the main branch.
- Writers run as background processes and execute code in the project workspace.
This is third-party code. Review the repository files before installing.
What it does
Claude Lane Stack is a plugin and toolkit that turns Claude Code into a project manager for several CLI agents. The manager writes each task as a contract, and writers (Codex, Qwen, Grok, Kimi, AGY, Cursor, OpenCode) run it as background processes in a git workspace. Every task carries its own `owns_paths` list. After the run come checks and an `acceptance.json` file, then the manager merges the result into `main`. A night shift sends commits to an independent Codex review, fixes findings and re-checks. State lives in files and git, with no cloud database. Work starts with `lane-pm` in the project directory.
Who it is for. A solo developer running several CLI agents in one project who wants review and merging without juggling windows.
Good fit when
- A feature is split into parts and each part is written by a separate CLI agent in its own paths.
- Daytime code goes to a night Codex review with fixes and a re-check.
- After a break you need to recall where you stopped: `resume-project` restores context from memory and git.
- Pages in `docs/` are stale and get refreshed from git history without manual edits.
Not a fit when
- A team already runs CI and human review: the conveyor merges into `main` by itself.
- Claude Code is not installed, so there is no project manager.
- The task is small and one-off, so setting up an agent profile does not pay off.
Example request
Add a dark theme to settings: hand the task to a writer, check the result and merge into main.Limitations
Claude Code is required; other CLI agents are optional. You need Git, Python 3 with PyYAML and jsonschema, Node, rsync and `flock`; on macOS install `flock` with `brew install flock`. The manager merges results into `main` after checks, so keep a backup of the repository. Some helper calls (context compaction, the router) go to external providers when their keys are set. Writer sandboxing on Linux needs `bubblewrap`, while macOS uses the built-in Seatbelt.
How to disable. Disable the `lane-stack` plugin in Claude Code, drop `~/.agents/bin` from `PATH` and delete `~/.agents`. Remove files that `agents-doctor` created in projects by hand.
FAQ
Which CLI agents are required?
Only Claude Code, which manages the project. Codex, Qwen, Grok, Kimi and AGY are optional; `agents-doctor` or `adoc` builds the profile.
How do I install only the plugin?
Run `claude plugin marketplace add VKirill/claude-lane-stack`, then `claude plugin install lane-stack@claude-lane-stack -y`. The `~/.agents` runtime must already be on the machine.
What happens to code after a task?
A writer works in `in_place`, `worktree` or `auto` mode. After checks and `acceptance.json` the manager merges the result into `main`.
Who reviews the code?
The night shift: an independent Codex pass over commits, then fixes and a re-check.
Is Codex required?
No. Codex adds the night review and can write code, but the conveyor also runs with Claude Code alone.
Discussion
No questions or reviews yet.
Related
PluginsClaude Code supervises while Codex, opencode and pi work on your quota: delegated edits, independent review, model discussions
Workflows for Claude Code: /do, /omo, /bmad-pilot and more commands, with Codex, Gemini or OpenCode writing the code through codeagent-wrapper
A plugin that syncs Claude Code, Codex CLI and Antigravity on one project and lets them hand work to each other.
A Claude Code plugin that hands coding tasks to Codex agents, monitors them and verifies the results independently
A template sets up a multi-agent environment in a repository: Claude Code conducts Sonnet and Opus subagents and calls Codex CLI.
A Claude Code skill that hands implementation-heavy work to Codex CLI while Claude keeps planning, review and acceptance.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Claude Lane Stack (this card) | no | 125 | |
| Lanes | no | 276 | Lanes is a task-board app for parallel agents; Lane Stack launches CLI writers itself and merges their work into `main` |
| AgentBridge: a bridge between Claude Code and Codex | no | 431 | agentbridge links Claude Code and Codex in one session; Lane Stack runs a conveyor with task contracts and a night review |
README in short
The repository is a Claude Code plugin marketplace and an installer for the `~/.agents` runtime. Normal work happens in chat with the manager, who prepares a plan, starts the controller and watches the run. Documentation and project memory are refreshed by separate jobs.