mirror of
https://github.com/multica-ai/multica.git
synced 2026-08-06 01:50:14 +02:00
e395fb8744
Complete rework of the docs site, in three passes squashed into one change: 1. P0 fact corrections across the core-loop pages (agent access model, comment triggers and coalescing, issue lifecycle, offline/queue semantics, task states and retries, dispatch via realtime push). 2. Chinese rewrite as the authoring language: final 12-group IA, new "Core concepts" and "Put agents to work" pages, curated welcome page, permanent-UI quickstart (no onboarding-wizard dependency), operator reference restored (CLI command reference, webhook response codes, task state/timeout tables, config set keys, upgrade notes). 3. Verification and sync: every load-bearing claim re-checked against current main with file:line evidence (migration 103 auto-backfill, CLI-only trigger toggling, duplication semantics, COOKIE_DOMAIN, GitHub App variables, Qwen Code as the 17th provider); register standards applied to all 39 pages; en/ja/ko synced from zh with same-language heading anchors (verified, zero dead links); 17 product screenshots plus 2 diagrams wired in. Supporting fixes: i18n middleware no longer swallows /images assets, sidebar folder indent survives custom link padding, unused editorial components removed. Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
96 lines
4.5 KiB
Plaintext
96 lines
4.5 KiB
Plaintext
---
|
|
title: Quickstart
|
|
description: Connect a computer, create your first agent, and have it complete its first task.
|
|
---
|
|
|
|
import { Callout } from "fumadocs-ui/components/callout";
|
|
|
|
<Callout type="info">
|
|
Before an agent can execute, the computer that runs it needs at least one [supported AI coding tool](/providers) installed and signed in — for example Claude Code, Codex, or Cursor. Multica does not ship these tools.
|
|
</Callout>
|
|
|
|
## 1. Sign in and open your workspace
|
|
|
|
Sign in on the web ([multica.ai](https://multica.ai)) or with [Multica Desktop](https://multica.ai/download). Email verification codes and Google are both supported.
|
|
|
|
- **Desktop**: can connect this computer automatically.
|
|
- **Web**: connect from the command line in the next step.
|
|
|
|
## 2. Connect a computer
|
|
|
|
Agents execute work on the computers you connect. Open **Configure → Runtimes** at the bottom of the sidebar:
|
|
|
|
- **With Desktop**: Desktop automatically registers this computer as a runtime and detects the installed AI coding tools. For a tool you just installed, click **Refresh** to detect it again.
|
|
- **With Web, or to add another computer**: click **Add a computer** in the top right, then run the two commands from the dialog in a terminal on the target computer:
|
|
|
|
**macOS / Linux**
|
|
|
|
```bash
|
|
curl -fsSL https://raw.githubusercontent.com/multica-ai/multica/main/scripts/install.sh | bash
|
|
multica setup
|
|
```
|
|
|
|
**Windows PowerShell**
|
|
|
|
```powershell
|
|
irm https://raw.githubusercontent.com/multica-ai/multica/main/scripts/install.ps1 | iex
|
|
multica setup
|
|
```
|
|
|
|
`multica setup` opens a browser to complete sign-in, then leaves the daemon running in the background. Once the daemon is online, the dialog usually detects the computer within a minute.
|
|
|
|
**Success check**: an online computer appears in the runtime list.
|
|
|
|
|
|
|
|
<Callout type="info">
|
|
The connection commands differ slightly on self-hosted instances — see the [Self-host quickstart](/self-host-quickstart).
|
|
</Callout>
|
|
|
|
## 3. Create an agent
|
|
|
|
Open **Workspace → Agents** in the sidebar, click **New agent**, and pick a path:
|
|
|
|
- **Start blank**: fill in the fields yourself. Only a name is strictly required — confirm the runtime and AI coding tool, then create.
|
|
- **Build with AI**: describe what you want, and the Agent Builder asks questions and generates the configuration. This path needs an online runtime.
|
|
|
|
Everything else can be refined later — see [Create and configure agents](/agents-create).
|
|
|
|
**Success check**: your agent appears in the agent list and shows as online.
|
|
|
|
|
|
|
|
## 4. Hand it your first issue
|
|
|
|
Click **New Issue** at the top of the sidebar (or press `C`). In the default agent mode:
|
|
|
|
1. Under **Created by**, select the agent you just created.
|
|
2. Describe the work in a sentence or two, for example "Explain how this workspace can be used."
|
|
3. Submit. Multica creates the issue, sets the agent as the assignee, and starts executing immediately.
|
|
|
|
|
|
|
|
## 5. Watch progress and results
|
|
|
|
After submitting, open the issue: the execution log shows the run status, and the agent's progress and replies appear in the timeline.
|
|
|
|
**Success check**: the execution log status changes to Completed and the agent's reply appears in the timeline. To see the full transcript, click **View transcript**.
|
|
|
|
|
|
|
|
## Common problems
|
|
|
|
| Symptom | Check first |
|
|
|---|---|
|
|
| No runtime found | Confirm the AI coding tool is installed and runs in a terminal. On Desktop, click **Refresh**; on the CLI path, rerun `multica setup`. |
|
|
| Runtime shows offline | Keep Desktop open; on the CLI path, run `multica daemon status`, and `multica daemon start` if it is not running. |
|
|
| Issue stays queued | Confirm the runtime the agent uses is online. When a runtime hits its concurrency limit, new tasks keep queueing. |
|
|
|
|
For more cases, see [Troubleshooting](/troubleshooting).
|
|
|
|
## Next steps
|
|
|
|
- [Core concepts](/concepts) — meet all of Multica's core objects in three minutes.
|
|
- [How Multica works](/how-multica-works) — how issues, agents, runtimes, and each run relate.
|
|
- [Put agents to work](/triggering-agents) — beyond assignment: @-mentions, chat, and Autopilots.
|