AGENT SKILL

models-and-configuration

Choose, configure, inspect, and troubleshoot CAMEL model backends, provider enums, config classes, local/OpenAI-compatible endpoints, ModelFactory, ModelManager, and structured output behavior.

License 许可
Apache-2.0
最近更新
2026-10-08
Stars
★ 11
Forks
3

01 前言 Front Matter

name
models-and-configuration
license
Apache 2.0
metadata
{ "disco-role": "operating" }
description
Choose, configure, inspect, and troubleshoot CAMEL model backends, provider enums, config classes, local/OpenAI-compatible endpoints, ModelFactory, ModelManager, and structured output behavior.
disable-model-invocation
true

02 资源文件 4

03 SKILL.md 正文

Models and Configuration

Use this sub-skill when a task involves CAMEL model selection, provider setup, model backend construction, model config dictionaries, structured outputs, local serving endpoints, or model fallback/scheduling.

Read This First

  • Start with API reference for ModelFactory, BaseModelBackend, ModelManager, ModelPlatformType, ModelType, audio model classes, and schema converters.
  • Use Provider patterns to choose OpenAI, OpenAI-compatible, Ollama, vLLM, LMStudio, cloud providers, custom clients, and credential handling patterns.
  • Use Configuration for camel.configs classes, model_config_dict, JSON/YAML factory configs, retries, timeouts, streaming, structured output settings, and optional extras.
  • Use Troubleshooting when imports fail, API keys are missing, provider/type pairs do not match, local endpoints refuse connections, structured output parsing fails, or Python 3.13 exposes optional dependency gaps.
  • Run inspect_model_registry.py to list installed CAMEL model platforms, backend classes, model enum counts, and config classes without making provider calls.

Common Entry Points

  • from camel.models import ModelFactory then ModelFactory.create(...) for most chat model backends.
  • from camel.types import ModelPlatformType, ModelType for provider/model enums; use a plain string model_type for local/custom model names that are not in ModelType.
  • from camel.configs import ChatGPTConfig, OllamaConfig, VLLMConfig, LMStudioConfig, ... and pass ConfigClass(...).as_dict() as model_config_dict.
  • Pass the model object, or a list of model objects, into camel.agents.ChatAgent(model=...); see sibling agents-and-societies for agent orchestration.
  • Keep embeddings and vector retrieval in sibling memory-rag-and-data, and tool schema/runtime details in sibling tools-runtimes-and-services.

Fast Routing

04 同仓库更多 Skills