🚀 MimikkAi Connect
MimikkAi — Быстрое создание AI-агентов под ваши задачи.
Welcome! Choose Your Own AI Adventure 🎯
Tired of being limited to a single AI model in Copilot Chat? Break free.
The MimikkAi Connect brings MimikkAi models directly into your VS Code Copilot Chat experience.
Whether you're experimenting with new models, optimizing costs, or running private LLMs behind a firewall—this extension lets you pick the right model without leaving your editor.
🛠️ Getting Started
- ✅ VS Code 1.120+ (required)
- 🌐 A MimikkAi account
- Install the extension from VS Code Marketplace
- Click MimikkAi: Connect in the status bar (or run MimikkAi: Connect Account from Command Palette)
- Choose your login method — Device Authorization (browser), Email & Password, or Attach Token
- Complete authorization — the extension auto-creates the provider group and
launches Setup automatically: confirm the prompts, and the required AI
settings are applied and the window reloads. (Re-configure anytime via
MimikkAi: Setup in the Command Palette or the status-bar menu.)
- Open Copilot Chat and pick a MimikkAi model
- Start chatting! 🎉
Status bar: 🟡 not connected / 🟢 connected. Click to toggle.
Hover over MimikkAi: Connected in the status bar to see your LiteLLM budget
window limits — one line per window with the percent used / percent left
(e.g. 6-часовой лимит · использовано 2% · осталось 98% · обновится в 12:00).
The set of windows is whatever your key has: hourly, weekly, both, or none.
Data is fetched from the LiteLLM proxy with your virtual key and cached for
24 hours (window percent refreshes every 5 minutes). If your key has no
budget windows, the tooltip simply shows the connection info.
🔌 Extension API (ai_api_token)
Other VSIX extensions can consume the MimikkAi tokens programmatically via
extension.exports (API version 1):
const ext = vscode.extensions.getExtension("MimikkAi.mimikkai-connect");
const api = ext?.isActive ? ext.exports : await ext?.activate();
const backendToken = await api.getMimikkaiToken(); // Sanctum token for service.mimikkai.ru
const litellmKey = await api.getLitellmKey(); // sk- key for litellm.mimikkai.ru
api.apiVersion === 1 — feature-detect before use.
- Both getters return
string | null (null when not connected) and never
show dialogs.
- Subscribe to
api.onDidChangeTokens — the LiteLLM key is rotated on every
reconnect; do not cache it across operations.
- ⚠️ Any installed extension can read these tokens. Install only extensions
you trust.
For manual inspection there are two commands in the Command Palette:
MimikkAi: Show Backend Token and MimikkAi: Show LiteLLM Key — both show
the value in a dialog with a Copy to Clipboard button.
🐛 Troubleshooting
| Issue |
Fix |
| Models not showing |
Run MimikkAi: Manage Configuration and check proxy status |
| Connection / timeout |
Verify proxy is running and network is reachable |
| Quota exceeded |
Check proxy rate limits; extension auto-redacts tools |