DATA CUT 2026-07-27 116 活跃模型 680 A 类案例

CASE EVIDENCE / A RECORD

dyagz 使用 Muse Spark 处理智能体流程编排

dyagz 公开的智能体工作流案例,来源为 公开代码库,复核于 2026-06-26T23:55:00Z。

原始记录:dyagz built a browser-backed LLM proxy exposing Muse Spark through OpenAI and Anthropic compatible APIs

A

Chinese Brief

中文案例导读

dyagz 公开的智能体工作流案例,来源为 公开代码库,复核于 2026-06-26T23:55:00Z。 Model Atlas 将它标记为 A 类证据,因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。 Model Atlas 不把 benchmark、教程、发布说明或集合页包装成真实案例。

厂商Meta / Llama
模型Muse Spark
任务类型智能体工作流
审核状态auto_approved

任务

真实任务背景

这是一个围绕智能体流程编排的真实任务,公开材料可以回溯到具体使用者和具体产物。 原始资料写作:Build a browser-backed proxy framework that exposes Meta AI's Muse Spark through both OpenAI-compatible /v1/chat/completions and Anthropic-compatible /v1/messages API frontends, with optional broker model integration fo…

智能体工作流公开代码库A 类可核验real_case
公开产物

公开材料提供公开代码、README 或项目配置,可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作:A Node.js proxy server with three-layer architecture: frontends (Anthropic + OpenAI format adapters), browser providers (meta.js for Meta AI session automation), and broker providers (zai.js for Z.ai GLM tool execution)…

模型作用

Muse Spark 在该案例中承担智能体流程编排相关的生成、分析、编排或实现角色。 原始资料写作:Muse Spark serves as the primary text generation backend accessed through the browser provider layer. The proxy automates browser interaction with Meta AI to provide structured API access to Muse Spark's capabilities fo…

风险边界

当前判断基于公开材料;若产物下线、仓库变更或模型参与比例仅来自作者自述,需要在引用前重新复核。 原始资料写作:Browser-backed integration is inherently fragile — UI selectors, auth flows, and anti-bot behavior can change without notice. Single-session only. Relies on undocumented Meta AI web interface.