---
title: "Alran Code: the AI agent that gets it done by lunch"
url: https://alran.dev/en
locale: en
updated: 2026-11-16
---

Alran Code

# Done by lunch.

The AI agent that plans, builds and tests your work. On Ottili Werk 1.1: fast, affordable, built for agents.

[Start free](https://ai.ottili.one)

`curl -fsSL https://alran.dev/install.sh | bash`

Welcome credit included · Open source (MIT) · Terminal and VS Code · 312 skills

## 38 tasks. One morning. {#h-film}

Give Alran Code a goal and it shares the work out to four agents. **Features, bugs, docs and the business tasks in between.** You check the result.

## Get the highlights. {#h-hl}

Plan first

### Say what you need. Alran reads the project and shows the plan before anything changes.

Working | fernhill-homes | Alran

```
● I'll add online booking to the listing pages. Plan: └ • Running   Todo list 0/6 · 6 remaining · Ctrl+T compact  ● Read the project  ○ Booking page  ○ Calendar with free times  ○ Save bookings  ○ Confirmation email  ○ Test everything
```

Goal Mode

### Set a goal. Alran keeps going, turn after turn, until it’s met.

Ready | fernhill-homes | Alran

```
● Done: every listing can be booked online.   └ Completed in 21min48s · ⚡ 210 tok/s   ✓ Goal complete · 1h20min0s · 1.1M tokens · 3 turns   ✓ Todo list 3/3 completed
```

Subagents

### One agent? Make it a team. Four workers, side by side.

Tasks | fernhill-homes | Alran

```
› ✓ [Agent] worker Done · Booking page · 0 tools · 10m  ✓ [Agent] worker Done · Calendar & free times · 0 tools · 11m  ✓ [Agent] worker Done · Confirmation emails · 0 tools · 12m  ✓ [Agent] worker Done · Tests · 0 tools · 13m
```

Checks its work

### Done means tested. The tests run before Alran calls the job finished.

Ready | fernhill-homes | Alran

```
├ • Wrote (db/migrations/004_bookings.sql)├ • Ran  Apply the bookings migration · 5 output lines├ • Wrote (src/email/booking-confirmation.mjs)├ • Wrote (test/booking.test.mjs)└ • Ran  Run all tests · 32 output lines   └ Completed in 15min39s · ⚡ 210 tok/s   ✓ Todo list 6/6 completed
```

Plugins & Skills

### Your tools, connected. Review, research, documents, spreadsheets, slides and PDFs built in. MCP for the rest.

- code-review
- deep-research
- docx
- xlsx
- pptx
- pdf
- skill-creator
- MCP

VS Code

### In your editor, too. VS Code, Cursor, Windsurf and VSCodium.

Ottili Werk 1.1

### Fast enough to keep an agent moving.

225tokens per second, up to

Open source

### MIT licensed. Read it, fork it, run it your way.

`github.com/Alran-AI/code`

Alran Code doesn’t autocomplete. It reads your project, plans the work, writes the code and runs the tests. You set the goal. It gets it done.

How it works

## From one sentence to working software. {#h-how}

1. **You say what you need.** One sentence. Alran reads the project first.

2. **Alran makes a plan.** A to-do list you see before anything changes.

3. **Alran builds it.** Code, commands, migrations, emails. Ticked off one by one.

4. **Alran checks it.** The tests run before it calls the job done.

## Built to finish. {#h-feat}

Everything an agent needs to see a job through. **And the controls to keep it yours.**

Sessions

### Pick up where you left off.

`alran --continue` brings back the last session, with its whole history.

Scripts and CI

### Runs without you watching.

`alran exec` runs headless in CI and scripts and streams its progress as JSON.

Permissions

### You decide what it may run.

Permission modes and a sandbox. Steps with effects outside your workspace wait for your OK.

Skills

## 312 skills, ready for your stack. {#h-skills}

**One command to install. No prompt engineering required.**

`alran plugin install stripe-skills`

- Cloudflare
- Supabase
- Stripe
- Firebase
- MongoDB
- Redis
- Neon
- Netlify
- Terraform
- Sentry
- Resend
- Expo
- Flutter
- React Native
- Hugging Face
- Gemini API
- ClickHouse
- DuckDB
- Cypress
- Sanity
- Replicate
- Apollo GraphQL
- and 7 more

[Browse the plugins and skills](https://alran.dev/en/docs/plugins)

Werk 1.1Ottili

## More than a coding model. {#h-werk}

Not an assistant. An agent. **Werk plans in steps, calls tools, reads what came back and finishes the job**, in your code and in the work around it.

Speed

225tok/s

Output speed, up to. An agent calls the model dozens of times per task; every second adds up.

225 tokens per second.

01

### Reasoning, tool use and agentic work

Thinks in steps, calls tools, recovers when a step fails.

02

### Coding

Reads the codebase, writes the change, runs the tests.

03

### Business tasks

Emails, documents, tables and research. The work around the code.

$0.50

per million input tokens

$1.50

per million output tokens

$0.03

per million cached tokens

[Meet Werk 1.1](https://alran.dev/en/werk)

VS Code

## Same agent. Right next to your code. {#h-vsc}

Plans, tool calls and approvals in a chat beside the editor. **Every change reviewable, every change reversible.**

[Alran Code for VS Code](https://alran.dev/en/vscode)

Install

## One line. Then one sentence. {#h-inst}

macOS, Linux and WSL. **The installer brings its own Node.js if yours doesn’t fit** and keeps everything in one folder.

1. **Install.** curl -fsSL https://alran.dev/install.sh | bash

2. **Sign in.** alran auth login

3. **Say what you need.** alran "Add online booking to our website."

Prefer npm? `npm install -g alran` · [All ways to install](https://alran.dev/en/download)

## Fits right in. {#h-tools}

Install it with the package manager you use, run it where you already work. **Your tools stay. Alran joins them.**

- Terminal
- VS Code
- Cursor
- Windsurf
- VSCodium
- Zed
- GitHub
- npm
- pnpm
- Yarn
- Bun
- Node.js
- TypeScript
- MCP

Names and logos belong to their owners and are shown to name compatible tools.

## Pick the model per task. {#h-models}

Werk runs on your credit, Claude on your own Anthropic key. **The welcome credit covers Werk.**

Default · in the welcome credit

### Ottili Werk 1.1

Fast, affordable, built for agents.

Input

$0.50

Output

$1.50

Cached input

$0.03

Longest answer

256K

Your Anthropic key

### Claude Sonnet 5.5

Anthropic’s model. Anthropic bills you directly.

Input

$2.00

Output

$10.00

Billed by

Anthropic

Your Anthropic key

### Claude Opus 5.5

Anthropic’s larger model, one switch away.

Input

$4.00

Output

$20.00

Billed by

Anthropic

Per one million tokens. Or bring your own provider key: those calls go straight to the provider.

## Why switch? {#h-why}

**4×**

### Less per input token

Werk 1.1 costs $0.50 per million input tokens. Claude Sonnet 5.5 lists at $2.

**Your key**

### Claude when you need it

Sonnet and Opus run on your own Anthropic key. Werk stays on your credit. Switch per task.

### Open, and your keys welcome

The code is on GitHub. Your own provider keys work too.

[Compare with Claude Code, Codex CLI and OpenCode](https://alran.dev/en/compare)

Powered by the Ottili AI Platform

## Put Werk into your product. {#h-api}

OpenAI-compatible. **Point your SDK at the platform and change the model name.** That’s it.

[Get an API key](https://ai.ottili.one/api-keys) [Ottili AI Platform](https://ai.ottili.one)

```
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://ai.ottili.one/v1",
    api_key=os.environ["OTTILI_API_KEY"],
)
reply = client.chat.completions.create(
    model="ottili/werk-1.1",
    messages=[{"role": "user", "content": "Summarize this ticket."}],
)
```

To do · Monday3 / 4

- Fix the flaky date test

- Add online booking

- Answer the customer emails

- Try Alran

## One task left. {#h-close}

Start free on the Ottili AI Platform. Then one line in your terminal.

[Start free](https://ai.ottili.one)`curl -fsSL https://alran.dev/install.sh | bash`
