Skip to content

cline.md

Latest commit

 

History

History
29 lines (19 loc) · 1.06 KB

cline.md

File metadata and controls

29 lines (19 loc) · 1.06 KB

Cline

Cline is a VSCode coding agent that supports any OpenAI-compatible provider.

1. Start the server

afm mlx -m mlx-community/Qwen3-Coder-Next-4bit \
  --port 9999 --enable-prefix-caching --concurrent 4

2. Configure Cline

In VSCode → Cline settings → API Provider:

  • API Provider: OpenAI Compatible
  • Base URL: http://localhost:9999/v1
  • API Key: any value (afm ignores it)
  • Model ID: the same model you started afm with, e.g. mlx-community/Qwen3-Coder-Next-4bit

3. Use it

Cline will issue streaming chat completions with tools. Tool format is auto-detected from model_type.

Tips

  • Cline's system prompt is large (~5–10k tokens) and stable across turns. Run with --enable-prefix-caching for instant turn-2+ prefill.
  • For multiple Cline tabs, raise --concurrent N to match.
  • Cline aborts mid-turn often (user edits, retries). Mid-stream cancel on client disconnect is on the roadmap (Tier 1.4); until landed, expect a few seconds of GPU idle after abort.