Skip to content

VS Code Integration

You can use TULIP from VS Code through the Continue extension. Continue supports OpenAI-compatible providers, so it can talk to TULIP's code endpoint directly.

TU Delft Network Access Required

TULIP is only reachable from the TU Delft network. Off campus, connect via eduVPN before using Continue.

Prerequisites

Configure Continue

  1. Install the Continue extension in VS Code.
  2. Open the Continue panel.
  3. Click the gear icon and select Open Config file.
  4. Replace the contents of config.yaml with:
name: TULIP
version: 1.0.0
schema: v1
models:
  - name: TULIP
    provider: openai
    model: AUTODETECT
    apiKey: YOUR_TULIP_API_KEY
    apiBase: https://api.tulip.tudelft.nl/code/v1

context:
  - provider: code
  - provider: file
  - provider: diff
  - provider: terminal
  - provider: problems
# Optional: add these later if you want broader project retrieval.
# - provider: docs
# - provider: codebase
# - provider: folder

Replace YOUR_TULIP_API_KEY with your actual TULIP API key, then save the file. Continue reloads the configuration automatically; if the model selector does not update, reload the VS Code window.

Try It

Open a source file in VS Code, open the Continue chat, and send:

Explain the current file.

Continue sends the request to TULIP through:

https://api.tulip.tudelft.nl/code/v1

The AUTODETECT model value lets Continue discover the models available behind the TULIP code capability.

Keep chat and code use separate

Continue is intended for coding workflows, so this configuration uses the code capability. For general chat outside VS Code, use the examples in Chat & General Inference.

Troubleshooting

Note

If Continue cannot connect, first check that you are on the TU Delft network or connected through eduVPN.

Note

If model discovery fails, verify that apiBase is exactly https://api.tulip.tudelft.nl/code/v1.

Known limitation

Continue currently has a known issue around Jupyter notebooks on remote machines. See continuedev/continue#1463.