Wildcard AI / agents.json开放规范和平台，允许AI代理通过agents.json文件发现并调用API流程。

5.0 (6)

审阅者 Daniel Nikulshyn·更新 2026年5月

Open Source Developer Tools AI Agents API Specification Workflow Automation Integration

概览

Wildcard AI 维护了一个名为 agents.json 的开源规范，该规范描述了 AI 代理如何查找和调用 API 端点以及多步骤工作流。开发人员不再硬编码工具调用或依赖脆弱的提示工程，而是可以在 API 旁边发布一个 agents.json 文件，以便任何兼容的代理都能理解有哪些操作可用以及如何链接它们。该平台帮助团队编写、托管和测试这些规范，并为代理提供运行时工具，以解析 agents.json 并针对真实的 API 执行所描述的工作流。其目标是为 AI 代理做 OpenAPI 对传统 API 客户端所做的事情，使集成更加声明式和可重用。它非常适合构建智能代理应用的开发者、希望其服务具备代理就绪能力的 API 提供商，以及寻找标准化的替代方案以取代逐模型函数调用格式的团队。

主要功能

agents.json规范以描述API动作
Workflow定义以串联多个端点
Runtime库为代理侧发现和执行
作者和托管工具以agents.json文件
与现有REST API兼容和认证方案
开源社区和参考实现

价格

模型: Freemium
分类: AI Agent Development Frameworks
评分: 5.0 / 5 (6)

使用场景

使 SaaS API 对 AI 代理可用

在现有REST API旁边发布agents.json文件，使任何兼容AI代理都可以发现可用的动作并无需自定义集成代码调用它们

定义多步骤API流程

将链式端点调用描述为可重用的 agents.json流程，让代理按预期执行复杂序列而不是通过.prompt来推断

取代易折断.prompt形式工具使用

利用runtime工具库为代理提供声明性的、结构化的工具定义，减少由硬编码工具调用或.prompt设计漂移引起的故障

编写和托管代理规范

团队可以使用平台的编写和托管工具来创建、测试和提供 agents.json文件，类似于管理OpenAPI规范的传统客户端

优点 & 缺点

优点

开源并且中立的规范
减少依赖于易折断的.prompt形式工具
支持多步骤API流程，不仅是单个调用
熟悉的模式对于使用OpenAPI而来的团队
支持现有API

缺点

标准仍然在发展中，尚未普遍接受
要求API提供者认真地编写和维护规范
比专有函数调用工具要小的生态系统
不如专有工具便利

评测

5.0

6 个评分的平均值。

登录以留下评测。

Aaliyah Johnson

May 4, 2026

Use it every day

Honestly didn't expect to like it this much. Runtime libraries for agent-side discovery and execution is exactly what I needed, and supports multi-step API workflows, not just single calls. but I reach for it almost every day now and it just clicks.

Jamal Carter

Sep 19, 2025

Years in this space

I've evaluated a lot of these over the years. What stands out here is runtime libraries for agent-side discovery and execution — handled better than most — and familiar pattern for teams already using OpenAPI. Worth the time if this is your use case.

Carlos Mendoza

Sep 18, 2025

Does the job

Pretty happy overall. Agents.json specification for describing API actions just works and reduces reliance on fragile prompt-based tool use. but no dealbreakers — I'd recommend it to a friend without hesitating.

Hiroshi Tanaka

Sep 8, 2025

Years in this space

I've evaluated a lot of these over the years. What stands out here is hosting and authoring tools for agents.json files — handled better than most — and reduces reliance on fragile prompt-based tool use. Worth the time if this is your use case.

Leila Hassan

Aug 12, 2025

Years in this space

Ahmed Saleh

Jun 13, 2025

Does the job

Pretty happy overall. Agents.json specification for describing API actions just works and supports multi-step API workflows, not just single calls. Standard is still emerging and not widely adopted can be annoying, but no dealbreakers — I'd recommend it to a friend without hesitating.

问答

How is agents.json different from OpenAPI or standard function-calling?

agents.json is an open spec specifically for AI agents, describing not just individual endpoints but multi-step workflows that chain calls together. It follows a familiar declarative pattern like OpenAPI, but targets agent-side discovery and execution rather than traditional API clients.

What do I need to do as an API provider to support Wildcard AI?

You author and host an agents.json file alongside your existing REST API, describing available actions and workflows. Wildcard provides authoring, hosting, and testing tools, and the spec is compatible with existing REST endpoints and auth schemes, so you don't need to rebuild your API.

Is agents.json production-ready and widely supported?

The specification is open-source and vendor-neutral with reference implementations, but it's still emerging and has a smaller ecosystem than proprietary function-calling tools. It's best suited to developers comfortable adopting early standards for agentic applications.