Use Tech AI monkey in VS Code
Tech AI monkey is a completely independent AI assistant powered by its own dedicated GPU engine. You can connect VS Code directly to this agent using our Chat Completions endpoint, or download the CLI app on the Download page.
Dedicated Local GPU Engine — Completely Independent
Unlike external services that rely on third-party APIs with expiring quotas, Tech AI monkey runs its own dedicated model engine (tech-ai-monkey) on our server. In VS Code, keep "toolCalling": false and use it in Ask mode because the agent handles research and code execution directly on the server.
1Get an API key
Open Dashboard → API keys and create a key. It looks like tai_… and is shown to you exactly once — copy it straight into VS Code. Only its SHA-256 hash is stored, and you can revoke it at any time from the same page.
You need to be signed in. Spending goes against your normal token balance; the owner account is unlimited.
2Add the model in VS Code
- Open the chat model picker in the Chat view and choose Manage Language Models (the gear icon) — or run
Chat: Manage Language Modelsfrom the Command Palette. - Select Add Models, then Custom Endpoint from the provider list.
- Enter a group name (for example Tech AI monkey), then paste your
tai_…key when asked. - Pick the API type Chat Completions.
- VS Code opens
chatLanguageModels.json— replace its contents with the block below and save.
3Paste the configuration
[
{
"name": "Tech AI monkey",
"vendor": "customendpoint",
"apiKey": "${input:techAiKey}",
"apiType": "chat-completions",
"models": [
{
"id": "tech-ai-monkey",
"name": "Tech AI monkey",
"url": "https://tech-ai.tech/v1/chat/completions",
"toolCalling": false,
"vision": false,
"maxInputTokens": 128000,
"maxOutputTokens": 8192
}
]
}
]The key is referenced as ${input:techAiKey} so it lives in VS Code's secret storage instead of sitting in the file as plain text. VS Code asks for it the first time the model is used.
4Select it in chat
The model now appears in the model picker as Tech AI monkey. If it does not show up straight away, reload the window.
5Use it anywhere else
Nothing here is VS Code specific — any client that lets you set a base URL and an API key works, for example Cline, Continue, an SDK, or your own server.
curl https://tech-ai.tech/v1/models \
-H "Authorization: Bearer tai_…"curl https://tech-ai.tech/v1/chat/completions \
-H "Authorization: Bearer tai_…" \
-H "content-type: application/json" \
-d '{
"model": "tech-ai-monkey",
"messages": [{ "role": "user", "content": "hello" }],
"stream": false
}'Set "stream": true to receive Server-Sent Events in standard format:
curl -N https://tech-ai.tech/v1/chat/completions \
-H "Authorization: Bearer tai_…" \
-H "content-type: application/json" \
-d '{
"model": "tech-ai-monkey",
"messages": [{ "role": "user", "content": "hello" }],
"stream": true
}'Endpoint reference
| Method | Path | Purpose |
|---|---|---|
| GET | /v1/models | Lists the models you can call |
| POST | /v1/chat/completions | Chat Completions, with stream or without |
Authentication is always Authorization: Bearer tai_…. Standard error format: { "error": { "message": "…", "type": "…" } }. Status 401 means the key is wrong, revoked or expired; 402 means the account is out of tokens.
Troubleshooting
- The model is missing from the picker. Reload VS Code. In Agent mode the picker only lists models with tool calling, which is why the configuration sets
toolCalling: falseand why it belongs in Ask mode. - 401 on every request. The key was revoked or the whole
Authorizationheader was overridden by a customrequestHeadersentry. - Answers feel like they forget the thread. Each request is a fresh agent session — the full message list is replayed into it. Very long conversations get summarised by the client, not by us.
- Corporate Copilot policy. Copilot Business/Enterprise admins can disable “Bring Your Own Language Model Key in VS Code” organisation-wide; that setting wins over anything configured locally.
Your data
Requests go to this instance using your own key and are metered against your account exactly like the web chat and the CLI. What is stored — prompts, replies, retention and deletion — is described in the privacy policy. Revoking the key in Dashboard → API keys cuts access off immediately.