Cline

Route Cline traffic through Axemere Gateway. Cline makes heavy agentic use of AI, including multi-step tool calls, file edits, and terminal commands, so routing through Axemere gives you full visibility into token spend and policy enforcement per task.

Table of Contents

• Prerequisites
• How It Works
• Setup
• Verify Traffic Is Flowing
• Next Steps

Prerequisites

• Cline VS Code extension installed: install from the VS Code Marketplace
• A gateway key: created in Console → Gateway Keys
• A workload ID: from Console → Workloads

How It Works

Cline is configured through its own VS Code extension settings. Set the provider to OpenAI Compatible and point the base URL at Axemere Gateway; every request Cline makes (chat, tool calls, and all agentic steps) routes through the gateway's policy engine.

Privacy: Axemere never stores, logs, or accesses the content of AI prompts or responses. Only request metadata (model, token counts, latency, attribution) is recorded. See Execution Records for exactly what is and isn't kept.

Setup

1. Open the Cline sidebar in VS Code and click the Settings (gear) icon, or go to VS Code Settings and search Cline.
2. Set API Provider to OpenAI Compatible.
3. Set Base URL to your gateway's OpenAI proxy URL. Include the /v1 suffix:

https://us.gw.axemere.ai/proxy/openai/k/mvgc_k_your-key-here/w/your-workload/v1

4. Set API Key to any placeholder value; the gateway authenticates via the key in the URL.
5. Set Model to the model you want to use (e.g., claude-sonnet-4-6 or gpt-4o).

Replace mvgc_k_your-key-here and your-workload with your values from the console.

Why OpenAI Compatible? Cline's native Anthropic and OpenAI provider entries have a known issue where the Base URL field is missing in some versions. The OpenAI Compatible provider always exposes the Base URL field and routes cleanly through the gateway. Anthropic models work correctly through the OpenAI-compatible path.

Verify Traffic Is Flowing

1. Open the Cline chat panel and send any message.
2. Check Records in the Axemere console; the request should appear within a few seconds, attributed to your workload.

Next Steps