OpenClaw provider plugin for ainetcafe: Kimi K3 served from ainetcafe's own cluster at the model's native MXFP4 precision, OpenAI-compatible, $2.10 / $10.50 per million tokens, $0.30 cached input.

Install

openclaw plugins install npm:openclaw-ainetcafe-provider --force --accept-capabilities
# --force: npm is outside ClawHub review, OpenClaw asks you to confirm the source (this repo is the source)
export AINETCAFE_API_KEY=sk-...        # https://microquickjs.com → Token Management ($2 sign-up credit)
openclaw models list --provider ainetcafe --refresh
openclaw agent --local --model ainetcafe/Kimi-K3 --thinking low -m "Say OK"
openclaw gateway restart               # if a gateway is already running

Or let onboarding store the key:

openclaw onboard --non-interactive --accept-risk --skip-health --mode local \
  --auth-choice ainetcafe-api-key --ainetcafe-api-key "$AINETCAFE_API_KEY"

Set it as the default model:

{ agents: { defaults: { model: { primary: "ainetcafe/Kimi-K3" } } } }

What it registers

Provider id ainetcafe, base URL https://microquickjs.com/v1, API openai-completions
Model ainetcafe/Kimi-K3: reasoning on (low / high / max via /think or --thinking), text + image input, 256K context (1M on request), reasoning_content kept on assistant turns so multi-step tool loops replay cleanly