IOanyT Code
An in-editor coding agent for Visual Studio Code, powered by IOanyT's own
hosted AI inference (currently Qwen3-Coder-Next-FP8). Chat about your code,
let the agent read and edit files in your workspace, and run shell commands —
all without sending your code to a third-party AI provider.
Your conversations and source code stay on IOanyT infrastructure.
Features
- 💬 Side-panel chat plus full chat-as-editor-tab — drag and dock like any tab
- 🗂 Multiple concurrent chats with separate histories per workspace
- 🛠 Agent tools — the assistant can
read_file, write_file, edit_file,
list_dir, glob, grep_search, and run_bash inside your project
- ✅ Confirmation cards before any write or shell command (toggleable)
- ⏸ Stop button to interrupt streaming responses mid-token
- 🧠 Workspace memory — drop an
AGENT.md in your repo root and the agent
reads it as project-specific context
- 🔒 Encrypted token storage via VS Code's built-in
SecretStorage — your
API token never sits in a settings file or on disk in plaintext
Installation
From the VS Code Extensions sidebar, search for IOanyT Code and click
Install.
Or from the command line:
code --install-extension ioanytinnovations.ioanyt-agent
VS Code keeps the extension up to date automatically — no need to reinstall
on new releases.
Getting started
- Get your API token. Contact IOanyT (devops@ioanyt.com) to be issued an
inference token.
- Open the IOanyT Code panel — click the IOanyT Code icon in the
activity bar, or press Ctrl+Alt+I (Cmd+Alt+I on macOS).
- Run "IOanyT Code: Set API Token" from the command palette
(Ctrl+Shift+P) and paste your token. It's stored securely in VS Code's
SecretStorage.
- Start chatting. Ask questions about your code, request edits, or have
the agent run a command — it will ask before making changes.
Quick sanity check: run IOanyT Code: Test Connection from the command
palette to confirm the endpoint is reachable.
Commands
| Command |
Default keybinding |
| IOanyT Code: Open Chat |
Ctrl+Alt+I / Cmd+Alt+I |
| IOanyT Code: New Chat |
Ctrl+Alt+N / Cmd+Alt+N |
| IOanyT Code: Set API Token |
— |
| IOanyT Code: Clear API Token |
— |
| IOanyT Code: Test Connection |
— |
Configuration
All settings live under ioanytAgent.* in VS Code settings.
| Setting |
Default |
Description |
ioanytAgent.endpoint |
IOanyT TIR endpoint |
Base URL of the inference endpoint. The extension appends /v1/chat/completions. |
ioanytAgent.model |
Qwen/Qwen3-Coder-Next-FP8 |
Model name as exposed by the endpoint. |
ioanytAgent.maxTokens |
16384 |
Maximum tokens per individual response. The agent auto-continues across responses if the model truncates, so effective output can be much larger. |
ioanytAgent.temperature |
0.2 |
Sampling temperature. Lower = more deterministic, good for coding. |
ioanytAgent.confirmActions |
true |
Prompt before running write_file, edit_file, or run_bash. Disable for "yolo mode" — at your own risk. |
The API token is not a setting. Set it via the command palette
(IOanyT Code: Set API Token) so it lives in SecretStorage, not on disk.
Privacy
- Code and chat content are sent only to the configured
ioanytAgent.endpoint,
which by default points at IOanyT's own hosted inference. No third-party AI
provider is involved.
- No telemetry is sent by the extension.
- The API token is stored in VS Code's encrypted
SecretStorage.
Support
License
MIT © IOanyT Innovations
