Data packages

API Provider Dataset

2,777 API providers · 34 fields each · 83% carry a source URL

$11.99 one payment · later versions free Buy Free sample →

Not available in the EEA, the UK or Switzerland. Eligibility →

What you download

  • Markdown archives (ZH) 13.3 MB · built Oct 2, 2026
  • JSON archives (ZH) 13.9 MB · built Oct 2, 2026
  • Markdown archives (EN) 13.1 MB · built Oct 2, 2026
  • JSON archives (EN) 13.4 MB · built Oct 2, 2026

Every file is also available on its own direct link, not only inside the archive — an agent should not have to unzip anything.

A complete archive

Notion, exactly as it ships. Nothing removed.

# Notion — API Provider 档案

<!-- NOTICE:BEGIN v2026-10-01 -->
> 本档案的数据检索、分析与归纳由 AI agent 完成,不排除存在不完善或错漏之处。网络信息也会随时间变化,因此使用这些数据时请把这一不确定性计入。建议通过档案内提供的来源链接、并辅以额外的网络检索做进一步核实。若你发现本档案与事实存在差异,欢迎通过官网或购买后台的联系方式告知我们;我们也会定期对数据做复核与补充。
<!-- NOTICE:END -->

- **官网**:https://www.notion.com
- **文档**:https://developers.notion.com  
  ↳ 来源:https://developers.notion.com
- **slug**:`notion`

## B · 公司画像
- **定位**:Notion 是一体化 AI 工作空间(all-in-one AI workspace),把笔记、文档、知识库(wiki)、项目、任务与数据库整合到同一平台,并在其上提供 Notion AI、Custom Agents、AI Meeting Notes、Enterprise Search 等 AI 能力,让团队把日常工作与 agent 放在一起协作。官网自我定位为『团队与 agent 共同思考的地方』(Where teams and agents think together),强调 capture 知识、instant 找答案、automate 琐事;官网称其被 98% 的 Forbes Cloud 100 公司使用。公司总部位于旧金山,团队背景涵盖计算、历史、艺术、编程语言等。
- **API 整体作用**:Notion 的开发者平台以 REST API(base URL https://api.notion.com)为核心,让『connection』(集成)把工作空间与外部 SaaS、自动化脚本或自建工具打通,并用代码读写工作空间中的内容。API 支持读取/创建/更新几乎一切对象——pages、databases/data sources、users、comments、views、files、blocks,并可通过 webhook 订阅实时事件。围绕同一套鉴权还衍生出多条产品线:Agent APIs(与 Custom Agents 开会话)、Agent Skills API(下载存于 Notion 的 skill)、Notion MCP(让 MCP 客户端连接工作空间)、Workers(在 Notion 内跑自定义代码/agent tool/同步)、CLI,以及面向企业的 Admin API(合规、审计、legal hold、权限组、MCP/PAT 管理等)。
- **业务逻辑**:以订阅制 SaaS 为主:按成员(seat)收费的 Free / Plus / Business / Enterprise 套餐(月付或年付,年付最高省 20%),并按套餐解锁 AI、权限、安全与合规能力。AI 相关功能采用 credits 计费(Custom Agents 免费试用后 $10 / 1,000 monthly Notion credits;Workers 自 10 月 15 日起消耗 credits)。开发者生态通过公开 connection 与 Marketplace 扩展,API 本身不单独计价,而是随套餐提供。  
  ↳ 来源:https://www.notion.com/pricing
- **目标用户**:开发者、团队与企业。细分包括:想把外部工具接入 Notion 或做内部自动化的开发者(internal connection / PAT);面向众多 Notion 用户与工作空间发布 SaaS 产品或服务的第三方(public connection,OAuth);以及做企业级合规、审计与权限管理的组织(Admin API)。  
  ↳ 来源:https://developers.notion.com
- **行业**:data-analytics, productivity-software

## C · 能力形状
**API 面(按 baseURL 分组)**:
- `https://api.notion.com/v1`
  - **Notion Blocks API**
  - **Notion Comments API**
  - **Notion Databases API**
  - **Notion Pages API**
  - **Notion Search API**
  - **Notion Users API**

**能力标签**:Collaboration, Database, Ideas, Notes, Productivity, Project, T1, Task, Wiki, Workspace

> ⚠️ apis[] 6 条为按 tag 拆分的 spec 文件, 归入 1 个 base 面, 非独立产品清单(计数=spec 非产品)。

- **行动清单(web 核实)**:创建页面(POST /v1/pages,支持 Notion-flavored Markdown 或 children blocks);检索/更新/移动/删除(trash)页面;按 markdown 读取/更新页面内容;查询 data source、创建/检索/更新 database 与 data source、列出 data source 模板;追加/检索/更新/删除 block children;创建/更新/删除/检索/列出 comment;按标题搜索(POST /v1/search);列出/检索 user 及当前 token 的 bot user;创建/发送/完成/检索/列出 file upload;创建/检索/更新/删除/列出 view 及其 query;创建与查询 AI meeting note;管理 Custom Agent(query/retrieve/insights/status/credit limit/delete/batch)与 agent session(创建、查询、取消、事件);下载 Agent Skills 的 plugin/skill;订阅 webhook 事件(page/database/data source/comment/file upload/view 的创建、更新、删除等);Admin API(legal hold、permission group、user、workspace export、MCP 连接、PAT 管理、agent 治理与 credits 策略)。  
  ↳ 来源:https://developers.notion.com/llms.txt
- **可用场景**:1) 把外部系统数据同步进 Notion 数据库并保持更新(Workers Syncs);2) 用 agent 自动创建/更新页面、生成 AI meeting notes、整理评论与任务;3) 用户在安装 public connection 后,自动为其工作空间创建数据库、页面与视图;4) 通过 webhook 实时响应页面/数据库/评论变化,驱动自动化工作流;5) 企业用 Admin API 做合规审计、legal hold 与权限管理。  
  ↳ 来源:https://developers.notion.com/llms.txt
- **不适合场景**:不适合需要极低延迟或不受限流约束的高频调用(每 connection 每分钟 600/180 请求,另有 workspace 级共享限流);单次请求受大小限制(payload 最多 1000 个 block、500KB,rich text 文本/URL 上限 2000 字符等),不适合大文件或海量数据传输(大文件须走多部分上传);不适合作为通用计算引擎,或用于非 Notion 场景的存储与计算。  
  ↳ 来源:https://developers.notion.com/reference/request-limits

## D · 自主闸门
- **onboarding 信号**:self-serve · **public**:False
- **必须人做的环节**:需要 Notion 账号与工作空间;创建 token 时若出现 workspace picker 需先选工作空间;在 Business 与 Enterprise 套餐上 PAT 创建默认被限制,需工作空间 owner 在 Settings → Connections 中开启;创建 connection 时需选择安装范围(Any workspace 或 Selected workspaces only,创建后不可更改),public connection 还需最终用户在 OAuth 流程中通过 page picker 授权可访问的页面。  
  ↳ 来源:https://developers.notion.com/guides/get-started/quick-start
- **程序化自助开通**:个人/开发者可自助开通:在 Developer portal 打开 Personal access tokens → New token → 填名称并勾选 Notion API capability → Create token,拿到 token 后即可用 Bearer 头直接调用 API,无需 OAuth 或合同;internal connection 同样可自助创建。前提是已拥有 Notion 账号/工作空间,且部分套餐(Business/Enterprise)需管理员先开启 PAT 创建权限。  
  ↳ 来源:https://developers.notion.com/guides/get-started/quick-start
- **程序化付费**:未提供程序化付费通道。Notion 目前仅接受主流信用卡与借记卡,套餐按成员计费、在账单周期结算,文档未提及 API 下单或无需人工输卡的自助支付方式。  
  ↳ 来源:https://www.notion.com/pricing
- **ToS 是否禁 agent**:*未采集*

## E · 经济形状
- **定价信号**:freemium
- **结算方式**:订阅制为主:按成员(seat)计费的 Free / Plus / Business / Enterprise 套餐,可选月付或年付(年付最高省 20%),成员增减按比例(prorated)结算;叠加按量计费的 Notion credits(AI 功能 / Custom Agents / Workers);站点自定义域名等按域名单独收费。  
  ↳ 来源:https://www.notion.com/pricing
- **免费档**:有 Free 档:个人使用 pages & blocks 不限量,含基础 forms/sites、Notion Calendar、数据库(含 subtasks/dependencies/custom properties)等;2 人及以上成员共享的 Free 工作空间有 block 数量上限(达到后仍可读/编辑/整理已有内容,但不能新增 block)。Plus/Business/Enterprise 提供 Notion AI 等的试用与更高额度。API 本身无单独定价,随工作空间套餐提供。  
  ↳ 来源:https://www.notion.com/pricing
- **单位成本**:Custom Agents:免费试用后 $10 per 1,000 monthly Notion credits;套餐本身按成员(seat)计价(具体单价以 pricing 页为准);Workers 自 10 月 15 日起消耗 Notion credits。  
  ↳ 来源:https://www.notion.com/pricing
- **rate limit**:每 connection 每 60 秒窗口:Business/Enterprise 600 请求/分钟(约 10/秒),其它套餐 180 请求/分钟(约 3/秒);另有跨工作空间所有 connection 共享的 per-workspace 限流(随套餐扩展),因此单个 connection 未超限也可能被限。超限返回 HTTP 429 与 code "rate_limited"(附 Retry-After,per-connection 最多 60 秒,per-workspace 可能更长);过载返回 529 "service_overload"。另有限制:payload 最多 1000 个 block、500KB。  
  ↳ 来源:https://developers.notion.com/reference/request-limits

## F · 集成成熟度 & 风险
- **交付模型**:open-core · **callable_host**:True
- **surface 信号**:Security, RateLimits, AgenticAccess, OpenAPI, Documentation, JSONSchema
- **鉴权**:以 Bearer token 鉴权,支持三种认证模型:Internal connection(单工作空间静态 API token)、Public connection(OAuth 2.0,用户授权时选择页面)、Personal access token(用户级静态 bearer token)。请求头为 Authorization: Bearer <token>,并需带 Notion-Version(如 2026-03-11)。  
  ↳ 来源:https://developers.notion.com
- **SDK**:官方 JavaScript SDK:@notionhq/client(github.com/makenotion/notion-sdk-js),文档示例同时给出 cURL、HTTP、JavaScript 版本;另有 Notion CLI 与 Workers SDK(@notionhq/workers),并可用任意支持 HTTP 的语言/库调用。  
  ↳ 来源:https://developers.notion.com/reference/intro
- **OpenAPI spec**:https://developers.notion.com/openapi.json  
  ↳ 来源:https://developers.notion.com/llms.txt
- **sandbox**:*未采集*
- **可逆性**:多数操作为对工作空间内容的读写,副作用相对可控:页面/块可 update,删除页面为 trash(移入回收站);但创建 comment 会通知相关用户、订阅的 webhook 会触发外部系统、file upload 会占用存储,删除 Custom Agent、revoke token、delete view/comment 等操作具不可逆或外部可见后果,调用前需谨慎。  
  ↳ 来源:https://developers.notion.com/llms.txt

---
*溯源:核实于 2026-09-17*

<!-- NOTICE:BEGIN v2026-10-01 -->
## 声明
1. 本档案是基于公开信息的事实汇编。
2. 其中引用的片段,权利归原权利人所有。
3. 标注为「自称」的内容是该公司的自我陈述,不是经我们核实的结论。
4. 所涉商标归各自所有者所有。
<!-- NOTICE:END -->

What is in it

One archive per API provider, each in the same fixed schema. Every archive ships twice — once as Markdown for reading, once as JSON for parsing. Same content, same field names, no divergence.

The schema was frozen before collection started. That is the point: an archive for a one-person API and an archive for Stripe have the same shape, so you can load the whole set into a dataframe on day one.

The 34 fields

A · Identity — where the thing actually lives

FieldWhat it holds
slugStable id, also the filename
nameProvider name
official_websiteHomepage, trailing slash normalised
doc_urlDeveloper documentation entry point

B · Positioning — who the company is

FieldWhat it holds
company_positioningWhat the company is, at company level
api_overall_roleWhat this API does inside that business
business_logicHow the company makes money
target_usersWho the API is for
industriesIndustry tags

C · Capabilities — what the API can actually do

FieldWhat it holds
api_surfacesEndpoints grouped by base URL: [{base_url, capabilities:[{name}]}]
tag_hintsCapability-domain tags
surface_caveatA note when apis[] entries are spec files rather than separate products
action_listVerb-level actions you can actually call, verified on the web
fit_scenariosTwo to four concrete things an agent can build with it
unfit_scenariosWhere it will not work, so you stop early

D · Autonomy gate — can software use it without a human in the loop

FieldWhat it holds
onboarding_signalself-serve / open / approval
publicWhether the API is publicly available
human_stepsWhich steps a human must perform: signup, approval, contract
self_serve_activationWhether a key can be obtained programmatically
programmatic_paymentWhether it can be paid for without someone typing a card
tos_agent_allowedWhether the terms of service forbid bots or agents

E · Economics — what it costs

FieldWhat it holds
pricing_model_signalfreemium / paid / open
settlement_shapeSubscription, usage-based or one-off
free_tierWhether there is one, and how much
unit_costCost per call or per unit
rate_limitRate limits and quotas

F · Integration — how hard it is to wire in, and what can go wrong

FieldWhat it holds
delivery_modelsaas / self-hosted / open-core / library
callable_hostWhether you call their endpoint or run it yourself
surface_signalsWhich artefacts exist: OpenAPI, SDKs, code examples, security docs
auth_shapeAPI key, OAuth, or something else
sdkOfficial SDKs and their languages
openapi_spec_urlAddress of the OpenAPI spec
sandboxWhether a test sandbox exists
reversibilityWhether a call has irreversible side effects — real money, real email

Plus a _provenance block on every archive: the source records it was built from and the date it was last verified.

How the fields are filled

Fields that need checking are filled from a page we can link to, and the link is stored next to the value. Across a 500-archive sample, 83.7% of fields carry a source URL. The rest are stored as not_found.

not_found is a value, not an absence. The schema forbids filling a gap with prose.

Questions

Is this a live API? No. It is a download. A hosted API is planned; the datasets are files today.

Are there rate limits? None. You have the files.

What about updates? One payment, and later versions are free. The collection is still running — the archive count goes up, not down.

What language is it in? Chinese and English. Every archive ships in both, with the same schema and the same field names — all four files (Markdown and JSON, in each language) are included in the one payment.

Can I use it commercially? Yes, inside your own products and research. You may not resell the dataset itself.

← All data packages