DISCOVERY · CODE
Syniray Code
Your next idea starts here.
A space to think, write and improve code with artificial intelligence. From the first question to the last review, with your reasoning at the centre.
Explore the plansOverview
An agent that works inside your project
Syniray Code is an AI software engineering platform that turns multiple AI models into autonomous coding agents. It can understand entire codebases, plan and execute development tasks, modify files, run commands and tests, debug errors, and continuously iterate toward a working solution. Developers can use Syniray’s managed infrastructure, bring their own AI providers and API keys, or connect local models and hardware for greater control, privacy, and flexibility.
It works in a project of its own on Syniray's cloud, or in a folder on your computer that you authorise. You choose where the model comes from and how much it may do without asking. Syniray Code is included in Pro and Max.
What it does
From understanding a project to shipping a change
Understand the code
It reads the files, searches the project and explains its structure, its entry points and its risks before touching anything.
Plan the task
A task becomes a sequence of steps you can follow as the agent works, with a live account of what it is doing.
Write and change code
It creates and edits files, and every change appears as a real diff with the lines added and removed.
Test and debug
It runs commands and tests where you allow it, reads the errors and fixes what it finds.
Improve and refactor
It cleans up and restructures code while keeping what already works, and explains the cause before fixing a bug.
Continue where it stopped
A task that hits the time limit, the end of your credits or another named limit pauses — Continue picks it up without repeating what was done.
Where it runs
Syniray's cloud or your own computer
Where the agent runs and where the model comes from are two separate choices.
Syniray Cloud
The agent works in an isolated workspace that belongs to your project, on Syniray. Nothing to install.
- A Files panel with the project's files
- The Syniray Browser to open and read web pages
- Attach files and images to a task
- On Max, choose the reasoning effort
My computer
Install the Syniray connector, pair it with a one-time code and authorise a folder. The agent works only inside the folders you allow.
- Running commands is a separate permission you grant
- Use local models, your own key or Syniray's models
- Web search through the connector
- See your computer's files from the page
Models
Choose where the model comes from
Syniray models
The models of your plan, run by Syniray and billed in credits — Syniray 2.0, the Pro models and, on Max, the Max-only models.
Your own API key
Connect a key from OpenAI, Anthropic, Google, DeepSeek or an OpenAI-compatible endpoint. Its usage is billed by your provider, not in credits. The key is stored encrypted and never shown again.
Local models
On your computer, use models served by Ollama or by a local OpenAI-compatible server. They need the agent to run on your computer.
Tools
What the agent can use
Files
Read, search, create and edit the project's files.
Commands and tests
Run commands and test suites in the environment you allow.
Git
See the repository's state and work with it.
Syniray Browser
Open and read web pages while it works. What a page says is treated as information, never as an order.
Web search
Search the web from a session on your computer, with the sources shown in the session.
Project instructions
Write how the agent should work in a project — or keep a SYNIRAY.md file — and every new session starts with it.
Control and permissions
You decide what it may do
Ask before changes
Every change needs your approval before it happens.
Trusted workspace
Edits inside the project run without asking. Anything sensitive still asks you.
Always asked
Secrets and credentials, network access, installing packages, deleting files and anything outside the project always need your answer — whatever the policy.
Web content stays untrusted
After the agent reads a web page, instructions written on it are never followed on their own.
Your computers, your rules
Rename a computer, change its folders and permissions, or remove it — its access ends at once.
Sessions you manage
Rename, stop or delete a session. It remembers its recent turns, and shows the credits it has used.
Credits
Paid as it works, never beyond your balance
With Syniray's models, each model call reserves its cost before it starts and is charged for what it really used. When your credits run out, the task pauses and tells you so; after a top-up, Continue resumes it. A task has no fixed number of steps: it ends when the work is done, when it stops making progress, or at its time limit — always with the reason shown.
Use cases
Where it helps
Add a feature
Describe it, let the agent build it and review the diff.
Write tests
Ask for tests for a module and have them run.
Fix a bug
Find the cause, explain it, then fix it.
Explain a codebase
Get the structure, the entry points and the risks of a project you have just opened.
Questions
Frequently asked questions
Do I need to install anything?
Not for the cloud: the agent works on Syniray. To work on your own computer, you install the Syniray connector and pair it with a one-time code.
Can the agent access my whole computer?
No. It works only inside the folders you authorise, and running commands is a separate permission you grant. You can change or remove a computer's access at any time.
Which models can I use?
Syniray's models on your plan, models from your own API key (OpenAI, Anthropic, Google, DeepSeek or an OpenAI-compatible endpoint), or local models on your computer with Ollama or a local OpenAI-compatible server.
Does my own API key cost credits?
No — calls to the model are billed by your provider. Only a web search the agent runs is charged in credits.
How are credits charged with Syniray's models?
Call by call: the cost is reserved before each call and charged for what was used. If credits run out, the task pauses and Continue resumes it after a top-up.
Can I see changes before they happen?
Yes. With “Ask before changes” every change needs your approval, and every change the agent makes appears as a diff.
What is SYNIRAY.md?
A file of instructions for the project — how to work in it, which commands to use, what not to touch. The agent reads it at the start of every new session.
Is Syniray Code included in the Free plan?
No. Syniray Code is part of Pro and Max.