Skip to main content

@ddse/acm-llm

@ddse/acm-llm supplies OpenAI-compatible clients used by the planner, framework helper, and AI Coder experience.

Installation​

pnpm add @ddse/acm-llm

Available clients​

ClientDescription
createVLLMClient(model, baseUrl)Connects to a self-hosted vLLM server
createOllamaClient(model, baseUrl?)Connects to Ollama models
OpenAICompatClientBase client for custom OpenAI-compatible providers

Clients expose:

  • name() — Provider identifier (vllm, ollama, etc.)
  • generate() — Standard completion interface
  • generateWithTools() — Structured tool-call interface used by the planner
  • metadata() — Provider-specific hints for budgeting

Example​

import {createVLLMClient} from '@ddse/acm-llm';

const client = createVLLMClient('Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8', 'http://localhost:8001/v1');
const response = await client.generateWithTools!(messages, tools, {
  temperature: 0.1,
  maxTokens: 512
});

Usage tips​

  • Use low temperatures (0.0–0.2) for deterministic planning.
  • Configure maxTokens to match your provider quotas.
  • Combine with the BudgetManager in AI Coder to enforce spend ceilings.