platform-assistant

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Anypoint Platform Operations

Anypoint Platform 操作指南

Overview

概述

Navigate the Anypoint Platform developer portal to discover available APIs, agent skills, and extension schemas. The portal is the single source of truth for all public Anypoint Platform API specifications and agent-executable workflows. The portal URL is provided in the agent context preamble injected at the top of this file.

The portal exposes three machine-readable discovery files designed for AI agent consumption. Start with

llms.txt

for a quick inventory, read

AGENTS.md

for the full operating manual, and query

registry.json

for programmatic access to every document.

Important: All portal files are plain-text source documents (YAML, JSON, Markdown). Always retrieve them with raw HTTP tools (curl, direct GET requests) — never pass portal URLs through summarizer or browser-rendering tools, which will truncate structured content.

在 Anypoint Platform 开发者门户中浏览，以发现可用的 API、Agent Skill 和扩展 schema。该门户是所有公开 Anypoint Platform API 规范和可由 Agent 执行的工作流的唯一真实来源。门户 URL 会在注入到此文件顶部的 Agent 上下文前言中提供。

该门户提供三个专为 AI Agent 使用而设计的机器可读发现文件。先使用

llms.txt

快速盘点，阅读

AGENTS.md

了解完整操作手册，再查询

registry.json

以编程方式访问所有文档。

重要： 所有门户文件均为纯文本源文档（YAML、JSON、Markdown）。务必使用原始 HTTP 工具（curl、直接 GET 请求）获取它们——切勿将门户 URL 传给摘要工具或浏览器渲染工具，否则会截断结构化内容。

llms.txt -- Quick Inventory

llms.txt -- 快速盘点

Fetch

{portal-url}/llms.txt

for a lightweight, one-page summary following the llmstxt.org convention. It lists:

A link to
```
AGENTS.md
```
(the full agent entry point)
A link to
```
registry.json
```
(the machine-readable catalog)
Schema references for
```
x-origin
```
and JTBD extensions
Every public API with a one-line description
Every agent skill with a one-line description

Use

llms.txt

when you need a quick overview of what the portal offers before diving deeper.

获取

{portal-url}/llms.txt

，以获得遵循 llmstxt.org 规范的轻量级单页摘要。它列出：

```
AGENTS.md
```
的链接（完整的 Agent 入口点）
```
registry.json
```
的链接（机器可读目录）
```
x-origin
```
和 JTBD 扩展的 Schema 引用
每个公开 API 及其单行描述
每个 Agent Skill 及其单行描述

在深入探索之前需要快速概览门户提供的内容时，使用

llms.txt

。

AGENTS.md -- The Agent Operating Manual

AGENTS.md -- Agent 操作手册

Fetch

{portal-url}/AGENTS.md

for the complete reference designed specifically for AI agents. This is the most important file to read thoroughly. It covers:

Site structure -- where every file type lives (specs, skills, schemas)
Registry format -- field definitions for
```
registry.json
```
entries

URN scheme and resolution -- how to convert identifiers to fetchable URLs:

urn:api:{slug}

resolves to

{portal-url}/apis/{slug}/api.yaml

urn:skill:{slug}

resolves to

{portal-url}/skills/{slug}/SKILL.md

urn:schema:x-origin

resolves to

{portal-url}/schemas/x-origin.schema.json

API spec conventions --
```
operationId
```
format, authentication, server variables
x-origin extension -- how dynamic parameters reference other APIs
Skill format (JTBD) -- step structure, input types, output chaining
Common agent workflows -- step-by-step patterns for discovery, execution, and resolution

获取

{portal-url}/AGENTS.md

，以获得专为 AI Agent 设计的完整参考文档。这是最重要的文件，应完整阅读。它涵盖：

站点结构——每种文件类型所在的位置（规范、Skill、Schema）
Registry 格式——
```
registry.json
```
条目的字段定义

URN 方案与解析——如何将标识符转换为可获取的 URL：

urn:api:{slug}

解析为

{portal-url}/apis/{slug}/api.yaml

urn:skill:{slug}

解析为

{portal-url}/skills/{slug}/SKILL.md

urn:schema:x-origin

解析为

{portal-url}/schemas/x-origin.schema.json

API 规范约定——
```
operationId
```
格式、身份验证、服务器变量
x-origin 扩展——动态参数如何引用其他 API
Skill 格式（JTBD）——步骤结构、输入类型、输出链式传递
常见 Agent 工作流——用于发现、执行和解析的分步模式

registry.json -- Programmatic Catalog

registry.json -- 程序化目录

Fetch

{portal-url}/registry.json

for the complete document index as a flat JSON array. Each entry contains:

```
$id
```
-- URN identifier (e.g.,
```
urn:api:api-manager
```
)
```
kind
```
-- document type:
```
oas
```
(API spec),
```
agent-skill
```
,
```
json-schema
```
, or
```
schema-doc
```
```
slug
```
-- URL-safe short name
```
name
```
and
```
description
```
-- human-readable metadata
```
href
```
-- relative path to the raw source file
```
docs
```
-- relative path to the rendered HTML page
```
apis
```
-- (skills only) array of API slugs this skill references

Filter by

kind

to segment the catalog:

Find all APIs:
```
kind == "oas"
```
Find all skills:
```
kind == "agent-skill"
```
Find skills for a specific API:
```
kind == "agent-skill"
```
AND
```
apis
```
contains the target slug

Find extension schemas:

kind == "json-schema"

kind == "schema-doc"

获取

{portal-url}/registry.json

，以获得完整的文档索引，它是一个扁平的 JSON 数组。每个条目包含：

```
$id
```
——URN 标识符（例如
```
urn:api:api-manager
```
）
```
kind
```
——文档类型：
```
oas
```
（API 规范）、
```
agent-skill
```
、
```
json-schema
```
或
```
schema-doc
```
```
slug
```
——适合 URL 的短名称
```
name
```
和
```
description
```
——便于阅读的元数据
```
href
```
——原始源文件的相对路径
```
docs
```
——渲染后 HTML 页面的相对路径
```
apis
```
——（仅适用于 Skill）此 Skill 引用的 API slug 数组

按

kind

筛选目录：

查找所有 API：
```
kind == "oas"
```
查找所有 Skill：
```
kind == "agent-skill"
```
查找特定 API 的 Skill：
```
kind == "agent-skill"
```
且
```
apis
```
包含目标 slug

查找扩展 Schema：

kind == "json-schema"

或

kind == "schema-doc"

Understanding x-origin (Dynamic Parameters)

理解 x-origin（动态参数）

Many API parameters have valid values that come from another API's response. These parameters carry an

x-origin

annotation in the OpenAPI spec:

yaml

parameters:
  - name: environmentId
    in: path
    schema:
      type: string
    x-origin:
      - api: urn:api:access-management
        operation: listEnvironments
        values: "$.data[*].id"
        labels: "$.data[*].name"

To resolve an

x-origin

parameter:

Read the
```
x-origin
```
array on the parameter
For each source, resolve the API URN to fetch the referenced spec
Find and call the referenced
```
operation
```
(by
```
operationId
```
)
Apply the
```
values
```
JSONPath expression to extract valid identifiers
Optionally apply the
```
labels
```
JSONPath to get human-readable display names

The

values

field is always required. The

labels

field is optional -- when absent, display the value directly. Both can be single JSONPath strings or arrays of paths.

Full schema:

{portal-url}/schemas/x-origin.schema.json

Documentation:

{portal-url}/schemas/x-origin-schema.md

许多 API 参数的有效值来自另一个 API 的响应。这些参数在 OpenAPI 规范中带有

x-origin

注解：

yaml

parameters:
  - name: environmentId
    in: path
    schema:
      type: string
    x-origin:
      - api: urn:api:access-management
        operation: listEnvironments
        values: "$.data[*].id"
        labels: "$.data[*].name"

要解析

x-origin

参数：

读取参数上的
```
x-origin
```
数组
对每个来源，解析 API URN，以获取引用的规范
查找并调用被引用的
```
operation
```
（通过
```
operationId
```
）
应用
```
values
```
JSONPath 表达式，提取有效标识符
可选地应用
```
labels
```
JSONPath，获取便于阅读的显示名称

values

字段始终为必填。

labels

字段为可选——缺失时，直接显示该值。两者既可以是单个 JSONPath 字符串，也可以是路径数组。

完整 Schema：

{portal-url}/schemas/x-origin.schema.json

文档：

{portal-url}/schemas/x-origin-schema.md

Understanding JTBD Skills

理解 JTBD Skill

Skills are multi-step API workflows written in markdown. Each skill's

SKILL.md

contains:

YAML frontmatter with
```
name
```
(kebab-case) and
```
description
```
(includes trigger terms for agent matching)
Numbered steps (
```
## Step N: Title
```
) each containing a YAML code block with:
- ```
api
```
  -- URN of the target API (e.g.,
```
urn:api:api-manager
```
  )
- ```
operationId
```
  -- the operation to call
- ```
inputs
```
  -- parameter mappings with four possible types
- ```
outputs
```
  -- values to capture for subsequent steps

The four input types are:

Type	Syntax	Use when
From another API	`from.api` + `from.operation` + `from.field`	Value must be fetched from an API call
From previous step	`from.variable`	Using an output captured earlier in the workflow
Literal	`value: "..."`	Static constant known in advance
User-provided	`userProvided: true`	Agent must prompt the user or use context

Outputs chain between steps -- a value captured as

name: environmentApiId

in Step 1 can be referenced as

from.variable: environmentApiId

in Step 2.

Some skills define Execution Paths -- alternative routes through the steps depending on what the agent already has (e.g., skip authentication if already logged in).

Full schema:

{portal-url}/schemas/jtbd-schema.md

Template:

{portal-url}/schemas/jtbd-template.md

Skill 是以 Markdown 编写的多步骤 API 工作流。每个 Skill 的

SKILL.md

包含：

YAML frontmatter，其中包含
```
name
```
（kebab-case）和
```
description
```
（包含用于 Agent 匹配的触发词）
编号步骤（
```
## Step N: Title
```
），每一步都包含一个 YAML 代码块，其中包含：
- ```
api
```
  ——目标 API 的 URN（例如
```
urn:api:api-manager
```
  ）
- ```
operationId
```
  ——要调用的操作
- ```
inputs
```
  ——具有四种可能类型的参数映射
- ```
outputs
```
  ——为后续步骤捕获的值

四种输入类型如下：

类型	语法	使用场景
来自另一个 API	`from.api` + `from.operation` + `from.field`	值必须从 API 调用中获取
来自上一步	`from.variable`	使用工作流前面步骤捕获的输出
字面值	`value: "..."`	预先已知的静态常量
用户提供	`userProvided: true`	Agent 必须提示用户提供或使用上下文

输出会在步骤之间链式传递——在 Step 1 中作为

name: environmentApiId

捕获的值，可以在 Step 2 中通过

from.variable: environmentApiId

引用。

某些 Skill 定义了 Execution Paths——根据 Agent 已经具备的内容，在步骤中采用不同的替代路径（例如，如果已经登录，则跳过身份验证）。

完整 Schema：

{portal-url}/schemas/jtbd-schema.md

模板：

{portal-url}/schemas/jtbd-template.md

Fetching Strategy

获取策略

Portal files (

*.md

*.yaml

*.json

) are plain-text source documents, NOT rendered web pages. When retrieving them:

Use raw HTTP tools (curl, direct GET, file-read tools) — NOT browser-rendering or summarizer tools (e.g., WebFetch). Summarizers truncate structured content like JSON arrays, YAML code blocks, and tables.
Parse responses verbatim —
```
registry.json
```
must be parsed as JSON, API specs as YAML, and skills as Markdown. Do not summarize or abbreviate.

Fetch files individually when you need their content:

Registry:
```
GET {portal-url}/registry.json
```
API spec:
```
GET {portal-url}/apis/{slug}/api.yaml
```
Skill definition:
```
GET {portal-url}/skills/{slug}/SKILL.md
```
Agent guide:
```
GET {portal-url}/AGENTS.md
```

门户文件（

*.md

、

*.yaml

、

*.json

）是纯文本源文档，不是渲染后的网页。获取时：

使用原始 HTTP 工具（curl、直接 GET、文件读取工具），而不是浏览器渲染或摘要工具（例如 WebFetch）。摘要工具会截断 JSON 数组、YAML 代码块和表格等结构化内容。
逐字解析响应——必须将
```
registry.json
```
解析为 JSON，将 API 规范解析为 YAML，将 Skill 解析为 Markdown。不要总结或缩写。

需要内容时单独获取文件：

Registry：
```
GET {portal-url}/registry.json
```
API 规范：
```
GET {portal-url}/apis/{slug}/api.yaml
```
Skill 定义：
```
GET {portal-url}/skills/{slug}/SKILL.md
```
Agent 指南：
```
GET {portal-url}/AGENTS.md
```

Prerequisites

前置条件

Network access -- ability to make HTTP requests to the portal URL (see agent context preamble for the base URL)
Anypoint Platform credentials (for executing skills) -- Bearer token or OAuth2 client credentials for API calls
HTTP fetch capability -- ability to fetch and parse YAML, JSON, and Markdown responses directly (raw HTTP, not via summarizer)

网络访问——能够向门户 URL 发起 HTTP 请求（有关基 URL，请参阅 Agent 上下文前言）
Anypoint Platform 凭据（用于执行 Skill）——用于 API 调用的 Bearer token 或 OAuth2 client credentials
HTTP 获取能力——能够直接获取并解析 YAML、JSON 和 Markdown 响应（使用原始 HTTP，而不是摘要工具）

Tips and Best Practices

提示与最佳实践

Discovery Strategy

发现策略

Start with
registry.json
for programmatic discovery -- it is structured, filterable, and machine-parseable
Read
AGENTS.md
when you need to understand conventions, resolve ambiguity, or learn how the portal works
Use
llms.txt
for a quick, lightweight overview before committing to a deeper exploration
Always use raw HTTP tools (curl, direct GET) for portal files -- never use summarizer tools that pass content through an intermediate model

从
registry.json
开始进行程序化发现——它结构化、可筛选且可由机器解析
阅读
AGENTS.md
，以了解约定、消除歧义或学习门户的工作方式
**使用
```
llms.txt
```
**快速、轻量地概览门户，然后再进行深入探索
始终使用原始 HTTP 工具（curl、直接 GET）获取门户文件——切勿使用会让内容经过中间模型的摘要工具

Working with x-origin

使用 x-origin

Always check x-origin on parameters before hardcoding values -- dynamic enums change per organization and environment
Chain x-origin calls: one API's x-origin often references Access Management, which itself provides foundational IDs for other APIs
Cache
organizationId
and
environmentId
-- they are reused across nearly every API call and rarely change within a session
Read the full x-origin array -- some parameters offer multiple sources (e.g., different API endpoints for different contexts)

在硬编码值之前，始终检查参数上的 x-origin——动态枚举值会随组织和环境而变化
链式调用 x-origin：一个 API 的 x-origin 通常会引用 Access Management，而后者又为其他 API 提供基础 ID
缓存
organizationId
和
environmentId
——它们几乎会在每次 API 调用中重复使用，并且在一个会话内很少发生变化
读取完整的 x-origin 数组——某些参数会提供多个来源（例如，针对不同上下文使用不同的 API 端点）

Skill Execution

Skill 执行

Read the full skill before starting -- execution paths may let you skip steps if prerequisites are already met
Chain variables between steps -- outputs from one step become inputs to the next via
```
from.variable
```
Check the troubleshooting section of each skill when errors occur -- skills document known failure modes and solutions
Resolve user-provided inputs early -- gather all
```
userProvided: true
```
inputs before starting execution to avoid mid-workflow interruptions

开始前阅读完整的 Skill——执行路径可能允许在前置条件已满足时跳过某些步骤
在步骤之间链式传递变量——一个步骤的输出会通过
```
from.variable
```
成为下一步的输入
发生错误时，检查每个 Skill 的故障排除部分——Skill 会记录已知故障模式及其解决方案
尽早解析用户提供的输入——在开始执行前收集所有
```
userProvided: true
```
输入，避免工作流执行中断

Portal Navigation

门户导航

Raw files via
href
-- use this for machine consumption (OpenAPI YAML, SKILL.md, JSON Schema)
Rendered pages via
docs
-- use this when you need HTML documentation (useful for extracting formatted content)
URN resolution -- always resolve URNs through the portal URL, never hardcode file paths

通过
href
获取原始文件——供机器消费（OpenAPI YAML、SKILL.md、JSON Schema）
通过
docs
获取渲染页面——需要 HTML 文档时使用（适合提取格式化内容）
URN 解析——始终通过门户 URL 解析 URN，切勿硬编码文件路径

Troubleshooting

故障排除

Portal Files Not Accessible

无法访问门户文件

Symptoms: HTTP errors when fetching llms.txt, AGENTS.md, or registry.json

Possible causes:

Incorrect portal URL
Network connectivity issues
Portal is being redeployed

Solutions:

Verify the base URL matches the one in the agent context preamble at the top of this file
Check that the full URL path is correct (no trailing slash issues)
Retry after a brief wait if the portal is temporarily unavailable

症状： 获取 llms.txt、AGENTS.md 或 registry.json 时出现 HTTP 错误

可能原因：

门户 URL 不正确
网络连接问题
门户正在重新部署

解决方案：

验证基 URL 与此文件顶部 Agent 上下文前言中的 URL 一致
检查完整 URL 路径是否正确（不存在尾部斜杠问题）
如果门户暂时不可用，请短暂等待后重试

Registry Returns Empty or Unexpected Results

Registry 返回空结果或意外结果

Symptoms: registry.json parses but filtering returns no results

Possible causes:

Using wrong
```
kind
```
value for filtering
API slug does not match any entry

Solutions:

Valid kind values are exactly:
```
oas
```
,
```
agent-skill
```
,
```
json-schema
```
,
```
schema-doc
```
List all slugs first to find exact matches (slugs are kebab-case, e.g.,
```
api-manager
```
not
```
API Manager
```
)
Skills have an
```
apis
```
array field -- APIs do not have a
```
skills
```
field

症状： registry.json 可以解析，但筛选后没有结果

可能原因：

筛选时使用了错误的
```
kind
```
值
API slug 与任何条目都不匹配

解决方案：

有效的 kind 值严格为：
```
oas
```
、
```
agent-skill
```
、
```
json-schema
```
、
```
schema-doc
```
首先列出所有 slug，以查找精确匹配项（slug 使用 kebab-case，例如
```
api-manager
```
，而不是
```
API Manager
```
）
Skill 包含
```
apis
```
数组字段，而 API 不包含
```
skills
```
字段

URN Resolution Fails

URN 解析失败

Symptoms: Fetching a URL constructed from a URN returns 404

Possible causes:

Incorrect URN-to-URL mapping
API or skill has been renamed or removed

Solutions:

Follow the exact resolution rules:

urn:api:{slug}

maps to

/apis/{slug}/api.yaml

, not

/apis/{slug}.yaml

Check
```
registry.json
```
for the
```
href
```
field which gives the canonical relative path
For schemas, resolution is not slug-based:
```
urn:schema:x-origin
```
maps to
```
/schemas/x-origin.schema.json
```

症状： 获取根据 URN 构建的 URL 时返回 404

可能原因：

URN 到 URL 的映射不正确
API 或 Skill 已被重命名或删除

解决方案：

遵循精确的解析规则：

urn:api:{slug}

映射到

/apis/{slug}/api.yaml

，而不是

/apis/{slug}.yaml

检查
```
registry.json
```
中的
```
href
```
字段，其中提供了规范的相对路径
对于 Schema，解析方式不基于 slug：
```
urn:schema:x-origin
```
映射到
```
/schemas/x-origin.schema.json
```

x-origin Operation Not Found

找不到 x-origin 操作

Symptoms: The operationId referenced in x-origin does not exist in the target API spec

Possible causes:

API spec version has changed
Incorrect operationId spelling (operationIds are camelCase)

Solutions:

Fetch the latest API spec via its URN and search for the operationId
Check the API spec's
```
paths
```
for all available operations

症状： x-origin 中引用的 operationId 在目标 API 规范中不存在

可能原因：

API 规范版本已更改
operationId 拼写错误（operationId 使用 camelCase）

解决方案：

通过 URN 获取最新的 API 规范，并搜索该 operationId
检查 API 规范的
```
paths
```
，查看所有可用操作

platform-assistant

Original

Translation

Anypoint Platform Operations

Anypoint Platform 操作指南

Overview

概述

llms.txt -- Quick Inventory

llms.txt -- 快速盘点

AGENTS.md -- The Agent Operating Manual

AGENTS.md -- Agent 操作手册

registry.json -- Programmatic Catalog

registry.json -- 程序化目录

Understanding x-origin (Dynamic Parameters)

理解 x-origin（动态参数）

Understanding JTBD Skills

理解 JTBD Skill

Fetching Strategy

获取策略

Prerequisites

前置条件

Tips and Best Practices

提示与最佳实践

Discovery Strategy

发现策略

Working with x-origin

使用 x-origin

Skill Execution

Skill 执行

Portal Navigation

门户导航

Troubleshooting

故障排除

Portal Files Not Accessible

无法访问门户文件

Registry Returns Empty or Unexpected Results

Registry 返回空结果或意外结果

URN Resolution Fails

URN 解析失败

x-origin Operation Not Found

找不到 x-origin 操作

Related Jobs

相关 Job