notion-api

作者: intellectronica

此技能提供透過REST呼叫與Notion API互動的完整說明。每當使用者要求與Notion互動時,包括讀取、建立、更新或刪除頁面、資料庫、區塊、評論或任何其他Notion內容,皆應使用此技能。技能涵蓋驗證、所有可用端點、分頁、錯誤處理及最佳實務。

npx skills add https://github.com/intellectronica/agent-skills --skill notion-api

Notion API Skill

This skill enables interaction with Notion workspaces through the Notion REST API. Use curl and jq for direct REST calls, or write ad-hoc scripts as appropriate for the task.

Authentication

API Key Handling

  1. Environment Variable: Check if NOTION_API_TOKEN is available in the environment
  2. User-Provided Key: If the user provides an API key in context, use that instead
  3. No Key Available: If neither is available, use AskUserQuestion (or equivalent) to request the API key from the user

IMPORTANT: Never display, log, or send NOTION_API_TOKEN anywhere except in the Authorization header. Confirm its existence, ask if missing, use it in requests—but never echo or expose it.

Request Headers

All requests require these headers:

-H "Authorization: Bearer $NOTION_API_TOKEN" \
-H "Notion-Version: 2025-09-03" \
-H "Content-Type: application/json"

Verifying Authentication

Test the API key by retrieving the bot user:

curl -s "https://api.notion.com/v1/users/me" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Base URL and Conventions

  • Base URL: https://api.notion.com
  • API Version: 2025-09-03 (required header)
  • Data Format: JSON for all request/response bodies
  • IDs: UUIDv4 format (dashes optional in requests)
  • Timestamps: ISO 8601 format (2020-08-12T02:12:33.231Z)
  • Property Names: snake_case
  • Empty Values: Use null instead of empty strings

Rate Limits

  • Average: 3 requests per second per integration
  • Bursts: Brief bursts above this limit are allowed
  • Rate Limited Response: HTTP 429 with Retry-After header
  • Strategy: Implement exponential backoff when receiving 429 responses

Request Size Limits

TypeLimit
Maximum block elements per payload1000
Maximum payload size500KB
Rich text content2000 characters
URLs2000 characters
Equations1000 characters
Email addresses200 characters
Phone numbers200 characters
Multi-select options100 items
Relations100 related pages
People mentions100 users
Block arrays per request100 elements

Confirmation for Destructive Operations

IMPORTANT: Before executing any operation that modifies or deletes data, ask the user for confirmation. This includes:

  • Updating pages or blocks
  • Deleting/archiving pages or blocks
  • Modifying database schemas
  • Creating pages (if multiple or in batch)
  • Any bulk operations

For a logical group of related operations, a single confirmation is sufficient.

Core API Endpoints

Search

Search across all accessible pages and databases:

curl -s -X POST "https://api.notion.com/v1/search" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "search term",
    "filter": {"property": "object", "value": "page"},
    "sort": {"direction": "descending", "timestamp": "last_edited_time"},
    "page_size": 100
  }' | jq

Filter values: "page" or "data_source" (or omit for both)

Pages

Retrieve a Page

curl -s "https://api.notion.com/v1/pages/{page_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Note: This returns page properties, not content. For content, use "Retrieve block children" with the page ID.

Create a Page

curl -s -X POST "https://api.notion.com/v1/pages" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "parent": {"page_id": "parent-page-id"},
    "properties": {
      "title": {
        "title": [{"text": {"content": "Page Title"}}]
      }
    },
    "children": [
      {
        "object": "block",
        "type": "paragraph",
        "paragraph": {
          "rich_text": [{"type": "text", "text": {"content": "Paragraph content"}}]
        }
      }
    ]
  }' | jq

Parent options:

  • {"page_id": "..."} - Create under a page
  • {"database_id": "..."} - Create in a database (legacy)
  • {"data_source_id": "..."} - Create in a data source (API v2025-09-03+)

Update a Page

curl -s -X PATCH "https://api.notion.com/v1/pages/{page_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "properties": {
      "title": {"title": [{"text": {"content": "Updated Title"}}]}
    },
    "icon": {"type": "emoji", "emoji": "📝"},
    "archived": false
  }' | jq

Additional update options: cover, is_locked, in_trash

Archive (Delete) a Page

curl -s -X PATCH "https://api.notion.com/v1/pages/{page_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{"archived": true}' | jq

Retrieve a Page Property Item

For properties with more than 25 references:

curl -s "https://api.notion.com/v1/pages/{page_id}/properties/{property_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Blocks (Page Content)

Retrieve Block Children

curl -s "https://api.notion.com/v1/blocks/{block_id}/children?page_size=100" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Use the page ID as block_id to get page content. Check has_children on each block for nested content.

Append Block Children

curl -s -X PATCH "https://api.notion.com/v1/blocks/{block_id}/children" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "children": [
      {
        "object": "block",
        "type": "heading_2",
        "heading_2": {
          "rich_text": [{"type": "text", "text": {"content": "New Section"}}]
        }
      },
      {
        "object": "block",
        "type": "paragraph",
        "paragraph": {
          "rich_text": [{"type": "text", "text": {"content": "Content here"}}]
        }
      }
    ]
  }' | jq

Maximum 100 blocks per request, up to 2 levels of nesting.

Position options in request body:

  • Default: appends to end
  • "position": {"type": "start"} - Insert at beginning
  • "position": {"type": "after_block", "after_block": {"id": "block-id"}} - Insert after specific block

Retrieve a Block

curl -s "https://api.notion.com/v1/blocks/{block_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Update a Block

curl -s -X PATCH "https://api.notion.com/v1/blocks/{block_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "paragraph": {
      "rich_text": [{"type": "text", "text": {"content": "Updated content"}}]
    }
  }' | jq

The update replaces the entire value for the specified field.

Delete a Block

curl -s -X DELETE "https://api.notion.com/v1/blocks/{block_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Moves block to trash (can be restored).

Databases

Retrieve a Database

curl -s "https://api.notion.com/v1/databases/{database_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Returns database structure including data sources and properties.

Query a Database

curl -s -X POST "https://api.notion.com/v1/databases/{database_id}/query" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "filter": {
      "property": "Status",
      "select": {"equals": "Done"}
    },
    "sorts": [
      {"property": "Created", "direction": "descending"}
    ],
    "page_size": 100
  }' | jq

See references/filters-and-sorts.md for comprehensive filter and sort documentation.

Create a Database

curl -s -X POST "https://api.notion.com/v1/databases" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "parent": {"page_id": "parent-page-id"},
    "title": [{"type": "text", "text": {"content": "My Database"}}],
    "is_inline": true,
    "initial_data_source": {
      "properties": {
        "Name": {"title": {}},
        "Status": {
          "select": {
            "options": [
              {"name": "To Do", "color": "red"},
              {"name": "In Progress", "color": "yellow"},
              {"name": "Done", "color": "green"}
            ]
          }
        },
        "Due Date": {"date": {}}
      }
    }
  }' | jq

Update a Database

curl -s -X PATCH "https://api.notion.com/v1/databases/{database_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "title": [{"text": {"content": "Updated Title"}}],
    "description": [{"text": {"content": "Database description"}}]
  }' | jq

Data Sources (API v2025-09-03+)

Data sources are individual tables within a database. As of API version 2025-09-03, databases can contain multiple data sources.

Create a Data Source

curl -s -X POST "https://api.notion.com/v1/data_sources" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "parent": {"type": "database_id", "database_id": "database-id"},
    "title": [{"type": "text", "text": {"content": "New Data Source"}}],
    "properties": {
      "Name": {"title": {}},
      "Description": {"rich_text": {}}
    }
  }' | jq

Users

List All Users

curl -s "https://api.notion.com/v1/users?page_size=100" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Retrieve a User

curl -s "https://api.notion.com/v1/users/{user_id}" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Retrieve Bot User (Self)

curl -s "https://api.notion.com/v1/users/me" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Comments

Retrieve Comments

curl -s "https://api.notion.com/v1/comments?block_id={block_id}&page_size=100" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" | jq

Use a page ID as block_id for page-level comments.

Create a Comment

On a page:

curl -s -X POST "https://api.notion.com/v1/comments" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "parent": {"page_id": "page-id"},
    "rich_text": [{"type": "text", "text": {"content": "Comment content"}}]
  }' | jq

Reply to a discussion:

curl -s -X POST "https://api.notion.com/v1/comments" \
  -H "Authorization: Bearer $NOTION_API_TOKEN" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "discussion_id": "discussion-id",
    "rich_text": [{"type": "text", "text": {"content": "Reply content"}}]
  }' | jq

Note: The API cannot start new inline discussion threads or edit/delete existing comments.

Pagination

Paginated endpoints return:

  • has_more: Boolean indicating more results exist
  • next_cursor: Cursor for the next page
  • results: Array of items

To iterate through all results:

  1. Make the initial request (omit start_cursor)
  2. Check has_more in the response
  3. If true, extract next_cursor and include it as start_cursor in the next request
  4. Repeat until has_more is false

Example request with cursor:

{
  "page_size": 100,
  "start_cursor": "v1%7C..."
}

Error Handling

HTTP StatusCodeDescription
400invalid_jsonRequest body is not valid JSON
400invalid_request_urlURL is malformed
400invalid_requestRequest is not supported
400validation_errorRequest body doesn't match expected schema
400missing_versionMissing Notion-Version header
401unauthorizedInvalid bearer token
403restricted_resourceToken lacks permission
404object_not_foundResource doesn't exist or not shared with integration
409conflict_errorData collision during transaction
429rate_limitedRate limit exceeded (check Retry-After header)
500internal_server_errorUnexpected server error
503service_unavailableNotion unavailable or 60s timeout exceeded
503database_connection_unavailableDatabase unresponsive
504gateway_timeoutRequest timeout

Best Practices

  1. Store IDs: When creating pages/databases, store the returned IDs for future updates
  2. Use Property IDs: Reference properties by ID rather than name for stability
  3. Batch Operations: Aggregate multiple small operations into fewer requests
  4. Respect Rate Limits: Implement exponential backoff for 429 responses
  5. Check has_more: Always handle pagination for list endpoints
  6. Validate Before Updates: Retrieve current state before making updates
  7. Use Environment Variables: Never hardcode API keys
  8. Handle Errors Gracefully: Check response status codes and error messages
  9. Schema Size: Keep database schemas under 50KB for optimal performance
  10. Properties Limit: Properties with >25 page references require separate retrieval

References

For detailed documentation on specific topics, see:

  • references/block-types.md - All supported block types and their structures
  • references/property-types.md - Database property types and value formats
  • references/filters-and-sorts.md - Database query filter and sort syntax
  • references/rich-text.md - Rich text object structure and annotations

相關技能

verify
facebook
針對React貢獻的預提交驗證,包含並行型別檢查與測試。依序執行格式化與語法檢查,於首次失敗時停止以提早發現問題。透過子代理程式並行執行型別檢查及雙重測試套件(原始碼與www),以提升效率。接受可選的測試模式參數,用以篩選執行的測試。當任何步驟失敗時,提供詳細的失敗報告與建議修正方式。
official
prompt-optimizer
sentry
建立、優化並反覆改進代理提示詞、系統提示詞、開發者提示詞及可重複使用的提示詞模板。當被要求改善某個…時使用。
official
mapbox-google-maps-migration
mapbox
為從 Google Maps Platform 遷移至 Mapbox GL JS 的開發者提供的遷移指南,涵蓋 API 對應、模式轉換及主要差異。
official
Search Company Knowledge
Atlassian
搜尋公司知識庫(Confluence、Jira、內部文件),以查找並解釋內部概念、流程與技術細節。當Claude需要:(1) 查找或搜尋關於系統、術語、流程、部署、驗證、基礎設施、架構或技術概念的資訊,(2) 搜尋內部文件、知識庫、公司文件或我們的文件,(3) 解釋某事物的定義、運作方式或查詢資訊,或(4) 綜合多個來源的資訊。會進行平行搜尋並提供附有引用的答案。
filesystem-agents
vercel
你是 Vercel Academy 上「Building Filesystem Agents」課程的專業教學助理,協助學生建立能透過 bash 操作檔案系統的代理,以回答關於結構化資料的問題。
official
speculative-decoding
firecrawl
Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques. Use when optimizing inference speed (1.5-3.6×…
official
bootstrap
vercel
Project bootstrapping orchestrator for repos that depend on Vercel-linked resources (databases, auth, and managed integrations). Use when setting up or…
official
encore-go-service
encoredev
Structure services with Encore Go.
official