Cline

Cline supports an OpenAI Compatible provider. Tokeness connects via the OpenAI-compatible endpoint — during configuration you fill in the Base URL, API Key, and Model ID.

Connection details

Field	Value
API Provider	OpenAI Compatible
Base URL	https://n.tokeness.dev/v1
API Key	A key created in the Tokeness console
Model ID	Full model name from the Tokeness model marketplace

If your Cline version does not offer OpenAI Compatible, choose OpenAI and then fill in a custom Base URL.

1. Prepare Tokeness

In the Tokeness console:

1. Confirm the wallet balance is available.
2. Create an API key dedicated to Cline.
3. Go to the model marketplace and copy a model name.
4. Open the usage logs — you will use them later to confirm requests.

You can name the key:

cline-local

Cline is a coding agent: it reads files, generates code, and retries tasks. Creating a separate key makes it easier to track consumption and troubleshoot issues later.

2. Open Cline settings

In VS Code, open the Cline panel and go to the settings page. The common entry point is the settings button in the top right of the Cline panel.

In the Provider or API Provider dropdown, select:

OpenAI Compatible

If the interface only shows an OpenAI option, check whether there is a Base URL, Custom Base URL, or API Base field. As long as you can change the Base URL, you can connect to Tokeness.

3. Fill in the Tokeness configuration

Enter the following:

Base URL: https://n.tokeness.dev/v1
API Key: your Tokeness API key
Model ID: the model name copied from the Tokeness model marketplace

Notes:

• Keep the /v1 suffix in the Base URL.
• Do not abbreviate the Model ID yourself.
• If Cline offers a Verify, Test, or Save button, verify once before saving.
• If the model list fetch fails, fill in the Model ID manually.

4. Verification

Send a read-only task in Cline:

Please read the current project structure, but do not modify any files. Summarize the top-level directories you see.

Check three things during verification:

1. Cline can reply normally.
2. A request appears in the Tokeness usage logs.
3. The key, model name, and status code in the logs are correct.

Do not let Cline make large-scale code changes on the first test. First confirm the connection, model, and permissions are all working, then proceed with file-writing tasks.

5. Settings before writing files

Cline performs file reads and modifications. After connecting to Tokeness, tool permissions are still controlled by Cline and VS Code — Tokeness only handles the model calls.

Check the following before writing files:

Item	Action
Key	Use a Cline-dedicated key
Quota	Set an appropriate quota on the key
Model	Choose a model suited for coding tasks
Logs	Keep Tokeness usage logs open and watch for repeated calls
Git	Confirm the working tree status before large changes

6. Troubleshooting

Symptom	Fix
Verify failed	Check the Base URL, key, and model name
401 Unauthorized	Re-copy the API key and confirm there are no spaces
404 Not Found	Base URL should be https://n.tokeness.dev/v1
model not found	Re-copy the Model ID from the Tokeness model marketplace
Cline reports errors but no Tokeness logs	The current provider is not routing through Tokeness
Replies work but tool calls fail	Switch to a model with stronger coding ability, or narrow the task scope
High consumption	Reduce the context range and use a separate key with a quota

External docs

• Cline OpenAI Compatible Provider
• Cline OpenAI Provider Custom Base URL