LEC-AI

Claude Devfleet — Git skill for Claude Code

Git community

Multi-agent coding platform that dispatches Claude CLI agents to work on missions (coding tasks).

How to install Claude Devfleet

This entry records only its repository, not the path inside it, so there is no exact command to give. Open LEC-AI/claude-devfleet and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Claude Devfleet does

Multi-agent coding platform that dispatches Claude CLI agents to work on missions (coding tasks). Each agent runs in an isolated git worktree, streams live output via SSE, and produces structured reports. Built on Claude Code's full feature set.

Alternatives in Git

Triage Open Pull Requests — Review, label, and act on all open PRs for a repository using parallel review agents 10.8k ★
Droidconkotlin — by touchlab - Delivers comprehensive Gradle commands for cross-platform Kotlin Multiplatform development with 1.1k ★
Fix GitHub Issue — by jeremymailen - Analyzes and fixes GitHub issues using a structured approach with GitHub CLI for issue detai 708 ★

README

Claude DevFleet — Autonomous Coding Agent Platform

[![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](LICENSE)

Multi-agent coding platform that dispatches [Claude Code](https://docs.anthropic.com/en/docs/claude-code) agents to work on missions (coding tasks). Agents run in isolated git worktrees with MCP-powered tools, create sub-missions for other agents, and coordinate autonomously through a dependency-aware dispatch system.

Built on the Claude Code SDK and MCP (Model Context Protocol) ecosystem.

Architecture

frontend/          React 19 + Vite UI (port 3101 via nginx)
backend/           FastAPI + SQLite (port 18801, uvicorn --reload)

Runs in Docker via `docker-compose.yml` or locally with a Python venv.

graph TD
    subgraph UI["Claude DevFleet UI :3101"]
        Web["React 19 + Vite"]
    end

    subgraph API["Claude DevFleet API :18801"]
        FastAPI["FastAPI + SQLite"]
    end

    subgraph Services["Background Services"]
        SDK["SDK Engine"]
        Watcher["Mission Watcher"]
        Sched["Scheduler"]
        AutoLoop["Auto-Loop"]
        Remote["Remote Control"]
    end

    subgraph Agents["Agent Pool (max 3)"]
        A1["Agent"] --> MCP1["MCP Servers"]
        A2["Agent"] --> MCP2["MCP Servers"]
        A3["Agent"] --> MCP3["MCP Servers"]
    end

    Web -->|"HTTP"| FastAPI
    FastAPI --> SDK
    FastAPI --> Watcher
    FastAPI --> Sched
    FastAPI --> AutoLoop
    FastAPI --> Remote
    SDK -->|"dispatch"| A1 & A2 & A3
    Watcher -->|"auto-dispatch on dependency satisfaction"| SDK
    Sched -->|"cron clone → dispatch"| Watcher
    AutoLoop -->|"plan → parallel dispatch"| SDK
    MCP1 -->|"create_sub_mission"| Watcher

Architecture Evolution

See the full [architecture evolution diagram](diagrams/devfleet-architecture-evolution.html) showing the platform's progression from CLI subprocess spawning (Phase 0) through SDK migration (Phase 1), MCP ecosystem (Phase 2), to autonomous multi-a