Quickstart
AdaL is the coding agent that does GTM directly from your codebase. What makes it stand out is multimodality — image, audio, and video, from understanding to generation to editing — built right into the agent.
It loads capabilities on its own: browser use, human-sounding writing, video, and deep research are there out of the box.
1. Install
macOS, Linux, WSL:
curl -fsSL https://adal.sylph.ai/install.sh | bash
Windows PowerShell:
irm https://adal.sylph.ai/install/windows | iex
The native installer manages AdaL's runtime and updates — no other prerequisites.
Prefer a window instead of a terminal? Download the Desktop app for macOS, Windows or Linux — same agent, plus a visual workspace for tasks and diff review.
2. Start AdaL
Open a terminal in your project folder and run:
adal
The first run opens your browser once to sign in; after that you stay logged in.
3. Say hi
Start with a conversation, not a manual:
Hi! What can you do for me?
AdaL will tell you about itself and suggest what to try next.
If you're in a codebase, ask for a tour:
Summarize this codebase — what does it do, and how do I run and test it?
From there, give it real work: "fix the failing test in @src/auth.test.ts" or "add rate limiting to the login endpoint". Prefix a file with @ to point AdaL at it. You approve every file edit and command before it runs.
4. Beyond code — AdaL's strength
Give AdaL one creative task and it chains every modality for you — images, video, and audio, from understanding to generation to editing. Ask for a video ad:
create a video ad: generate an image of a pink electric sports car on a coastal
road at sunset, use that image as the reference first frame for a 5-second video,
write a 2-line voiceover script and generate the voiceover, then composite the
video and voiceover into one final video
The video capability loads on its own when the task needs it — no plugins to install, no setup. You get the generated image, the voiceover script and audio, and one final composited video. To find the video, watch it on its timeline and compare its versions, see Video Producer.
5. Try a capability
Capabilities and skills load out of the box — when a task needs one, AdaL brings it in itself. Try pointing it at a running app, or ask it what it can do:
open my app at localhost:3000, click the login button, and screenshot what happens
Browse everything available with /skills.
6. Everyday commands
| Input | What it does |
|---|---|
/model | Switch AI model (10+ providers) |
/usage | Check your credit usage and limits |
/init | Generate project context (AGENTS.md) |
/resume | Resume a previous conversation |
/help | Show all commands |
? | Show all shortcuts |
@ | Attach a document or image as context — @report.pdf summarize this |
Sessions save automatically — quit anytime and resume later.
Where to go next
- Video Producer — ask for a video in plain words, then watch it, compare its versions and see every clip it used
- Capabilities — what AdaL can do, and how it loads them
- Skills & Plugins — extend AdaL with skills and MCP servers
- Models & Billing — every model, credits, and usage
- Terminal Guide — recommended terminals and truecolor setup