The open-source managed agents platform. Turn coding agents into real teammates — assign tasks, track progress, compound skills. https://multica.ai
Find a file
LinYushen 6d2a0b45d2
refactor: decouple task lifecycle from issue status (#151)
* refactor: decouple task lifecycle from issue status, add daemon health server

- Remove automatic issue status changes from StartTask (in_progress),
  CompleteTask (in_review), and FailTask (blocked) in task service.
  Issue status is now fully managed by the agent via `multica issue status`.
- Update agent prompt and meta skill to instruct agents to manage issue
  status themselves (in_progress → done/in_review/blocked).
- Add daemon health HTTP server on 127.0.0.1:19514 with /health endpoint
  exposing pid, uptime, agents, and workspaces. Fail fast if port is taken
  (another daemon already running).
- Update `multica status` to check both server and daemon health.
- Add Save button to repos section in workspace settings UI.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* refactor(daemon): simplify prompt, fix runtime config path, improve task error logging

- Slim down BuildPrompt to a minimal hint; detailed workflow now lives in CLAUDE.md/AGENTS.md
- Write CLAUDE.md to workDir root instead of .claude/CLAUDE.md
- Fix git-exclude pattern (.claude → CLAUDE.md)
- Decouple task queue reconciliation from issue status changes (agents manage status via CLI)
- Add diagnostic logging when CompleteTask/FailTask fail due to unexpected task state

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix(task): use task_completed/task_failed inbox notification types

FailTask was sending "agent_blocked" which conflates agent crash with
issue-level blocked status. Align notification types with the new
decoupled model: task_completed and task_failed. Update frontend types
and labels accordingly.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-27 18:30:21 +08:00
.github/workflows fix(ci): replace turbo commands with direct pnpm calls 2026-03-26 18:01:24 +08:00
apps/web refactor: decouple task lifecycle from issue status (#151) 2026-03-27 18:30:21 +08:00
e2e fix: resolve merge conflicts with main, preserve PAT functionality 2026-03-26 17:19:24 +08:00
scripts refactor(dev): share postgres across main and worktrees 2026-03-24 14:27:35 +08:00
server refactor: decouple task lifecycle from issue status (#151) 2026-03-27 18:30:21 +08:00
.env.example feat(auth): email verification login and personal access tokens 2026-03-26 14:32:30 +08:00
.gitignore refactor: remove unused monorepo tooling 2026-03-26 17:57:11 +08:00
.goreleaser.yml chore: rename super-multica references to multica 2026-03-25 21:01:39 +08:00
.npmrc refactor: restructure to monorepo architecture 2026-02-10 18:00:23 +08:00
AGENTS.md refactor: decouple task lifecycle from issue status (#151) 2026-03-27 18:30:21 +08:00
CLAUDE.md docs: update CLAUDE.md with logging, CI, and CLI details 2026-03-27 12:33:38 +08:00
docker-compose.yml chore: rename super-multica references to multica 2026-03-25 21:01:39 +08:00
LOCAL_DEVELOPMENT.md chore: rename super-multica references to multica 2026-03-25 21:01:39 +08:00
Makefile chore(web): remove knowledge base module and fix worktree port conflicts 2026-03-26 20:57:14 +08:00
package.json refactor: remove unused monorepo tooling 2026-03-26 17:57:11 +08:00
playwright.config.ts Add workspace management and isolated worktree environments 2026-03-23 18:12:11 +08:00
pnpm-lock.yaml feat: settings redesign, rich text mentions, inbox listeners, and UI polish 2026-03-26 19:17:14 +08:00
pnpm-workspace.yaml refactor: remove unused monorepo tooling 2026-03-26 17:57:11 +08:00
README.md refactor(cli): move watch/unwatch under workspace subcommand 2026-03-26 18:28:44 +08:00
skills-lock.json feat: add Claude Code skills for UI/UX development 2026-03-27 00:40:16 +08:00

Multica

AI-native task management platform — like Linear, but with AI agents as first-class citizens.

For the full local development workflow, see Local Development Guide.

Prerequisites

Quick Start

# 1. Install dependencies
pnpm install

# 2. Copy environment variables for the shared main environment
cp .env.example .env

# 3. One-time setup: ensure shared PostgreSQL, create the app DB, run migrations
make setup

# 4. Start backend + frontend
make start

Open your configured FRONTEND_ORIGIN in the browser. By default that is http://localhost:3000.

Main checkout uses .env. A Git worktree should generate its own .env.worktree and use the explicit worktree targets:

make worktree-env
make setup-worktree
make start-worktree

Every checkout shares the same PostgreSQL container on localhost:5432. Isolation now happens at the database level:

  • .env typically uses POSTGRES_DB=multica
  • each .env.worktree gets its own POSTGRES_DB, such as multica_my_feature_702
  • backend/frontend ports still stay unique per worktree

That keeps one Docker container and one volume, while still isolating schema and data per worktree.

Project Structure

├── server/             # Go backend (Chi + sqlc + gorilla/websocket)
│   ├── cmd/            # server, daemon, migrate
│   ├── internal/       # Core business logic
│   ├── migrations/     # SQL migrations
│   └── sqlc.yaml       # sqlc config
├── apps/
│   └── web/            # Next.js 16 frontend
├── packages/           # Shared TypeScript packages
│   ├── ui/             # Component library (shadcn/ui + Radix)
│   ├── types/          # Shared type definitions
│   ├── sdk/            # API client SDK
│   ├── store/          # State management
│   ├── hooks/          # Shared React hooks
│   └── utils/          # Utility functions
├── Makefile            # Backend commands
├── docker-compose.yml  # PostgreSQL + pgvector
└── .env.example        # Environment variable template

Commands

Frontend

Command Description
pnpm dev:web Start Next.js dev server (uses FRONTEND_PORT, default 3000)
pnpm build Build all TypeScript packages
pnpm typecheck Run TypeScript type checking
pnpm test Run TypeScript tests

Backend

Command Description
make dev Run Go server (uses PORT, default 8080)
make daemon Run local agent daemon
make test Run Go tests
make build Build server & daemon binaries
make sqlc Regenerate sqlc code from SQL

Database

Command Description
make db-up Start the shared PostgreSQL container
make db-down Stop the shared PostgreSQL container
make migrate-up Ensure the current DB exists, then run migrations
make migrate-down Rollback database migrations for the current DB
make worktree-env Generate an isolated .env.worktree for the current worktree
make setup-main / make start-main Force use of the shared main .env
make setup-worktree / make start-worktree Force use of isolated .env.worktree

CLI (multica)

The CLI manages authentication, workspace configuration, and the local agent daemon.

Install

brew tap multica-ai/tap
brew install multica-cli

Or build from source:

make build
cp server/bin/multica /usr/local/bin/multica   # or ~/.local/bin/multica

Authentication

multica auth login          # Open browser to authenticate (one-click if already logged in)
multica auth login --token  # Paste a personal access token manually
multica auth status         # Show current auth status
multica auth logout         # Remove stored token

Credentials are saved to ~/.multica/config.json.

Workspaces

multica workspace list      # List all workspaces you belong to

Daemon Watch List

The daemon monitors one or more workspaces for tasks. Manage which workspaces are watched:

multica workspace watch <workspace-id>    # Add a workspace to the watch list
multica workspace unwatch <workspace-id>  # Remove a workspace from the watch list
multica workspace list                    # Show all workspaces (watched ones marked with *)

The watch list is stored in ~/.multica/config.json. Changes are picked up by a running daemon within 5 seconds (hot-reload).

Local Agent Daemon

The daemon polls watched workspaces for tasks and executes them using locally installed AI agents (Claude Code, Codex).

# 1. Authenticate
multica auth login

# 2. Add workspaces to watch
multica workspace watch <workspace-id>

# 3. Start the daemon
multica daemon

The daemon auto-detects available agent CLIs (claude, codex) on your PATH. When a task is claimed, it creates an isolated execution environment, runs the agent, and reports results back to the server.

Other Commands

multica agent list          # List agents in the current workspace
multica runtime list        # List registered runtimes
multica config show         # Show CLI configuration
multica version             # Show CLI version

Environment Variables

See .env.example for all available variables:

  • DATABASE_URL — PostgreSQL connection string
  • POSTGRES_DB — Database name for the current checkout or worktree
  • POSTGRES_PORT — Shared PostgreSQL host port (fixed to 5432)
  • PORT — Backend server port (default: 8080)
  • FRONTEND_PORT / FRONTEND_ORIGIN — Frontend port and browser origin
  • JWT_SECRET — JWT signing secret
  • MULTICA_APP_URL — Browser origin for CLI login callback (default: http://localhost:3000)
  • MULTICA_DAEMON_ID / MULTICA_DAEMON_DEVICE_NAME — Stable daemon identity for runtime registration
  • MULTICA_CLAUDE_PATH / MULTICA_CLAUDE_MODEL — Claude Code executable and optional model override
  • MULTICA_CODEX_PATH / MULTICA_CODEX_MODEL — Codex executable and optional model override
  • MULTICA_WORKSPACES_ROOT — Base directory for agent execution environments (default: ~/multica_workspaces)
  • NEXT_PUBLIC_API_URL — Frontend → backend API URL
  • NEXT_PUBLIC_WS_URL — Frontend → backend WebSocket URL

Local Development Notes

  • make setup, make start, make dev, and make test now require an env file. They fail fast if .env or .env.worktree is missing.
  • make stop only stops the backend/frontend processes for the current checkout. It does not stop the shared PostgreSQL container.
  • Use make db-down only when you explicitly want to shut down the shared local PostgreSQL instance for every checkout.