Connect and work with coding agents
Vortex brings your connected coding agent into the project conversation. Your provider account supplies model access and its allowance; Vortex supplies the surrounding project workflow, preview and tools. Start by confirming installation, sign-in and one successful project run.
Connect the provider you already use
Open Agents & chat in Settings, inspect installation status and complete the supported account flow. An executable being present is different from a verified account. Choose the provider in a new conversation and ask it to inspect the selected project before requesting edits.
Guide a run without losing the thread
Choose the access mode for the work, answer approvals in the owning run and send a bounded follow-up when requirements change. A provider can accept steering during the active turn or retain it for the next one. Read the queue state before repeating an uncertain instruction.
Keep the account limits understandable
Conversation context, external provider quota and cloud storage are different measures. Models and usage reports vary between providers. Missing quota information is not unlimited usage, and Vortex does not include an AI-token allowance.
Follow the detailed guides
- Connect Codex
Connect Codex once, then choose it in a project conversation. Your Vortex sign-in and provider sign-in are separate accounts.
- Connect Claude Code
Connect Claude Code once, then choose it in a project conversation. Your Vortex sign-in and provider sign-in are separate accounts.
- Connect Google Antigravity
Connect Google Antigravity once, then choose it in a project conversation. Your Vortex sign-in and provider sign-in are separate accounts.
- Connect Grok Build
Connect Grok Build once, then choose it in a project conversation. Your Vortex sign-in and provider sign-in are separate accounts.
- Connect Kimi Code
Connect Kimi Code once, then choose it in a project conversation. Your Vortex sign-in and provider sign-in are separate accounts.
- Choose what your agent can do
Choose how much independence an agent should have for this conversation. For a first project, supervised work gives you a chance to review unfamiliar steps as they arise.
- Add an instruction while work continues
You can add a detail while your agent is working. Keep the new instruction specific enough that the agent can tell whether it changes the current task or comes next.
- Choose a model and read its usage
Choose a model for the current conversation after connecting its provider. Defaults help you start the next chat; changing a default does not rewrite the model choices in existing conversations.
- Show the agent what you mean
A screenshot or sketch can explain a design change better than a paragraph. Attach it to the draft or conversation, then say which part the agent should use.
- Speak a message
Speak the next change when it is easier than typing. Ordinary dictation puts recognized text in the draft so you can review it before sending.