langchain - 💡(How to fix) Fix Add NotionSearchTool and NotionGetPageTool to langchain-community [1 participants]

Submission checklist

• This is a feature request, not a bug report or usage question.
• I added a clear and descriptive title that summarizes the feature request.
• I used the GitHub search to find a similar feature request and didn't find it.
• I checked the LangChain documentation and API reference to see if this feature already exists.
• This is not related to the langchain-community package.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-openrouter
• langchain-perplexity
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Feature Description

Motivation

LangChain currently has Notion document loaders (NotionDBLoader, NotionDirectoryLoader) but no native Notion tools in langchain-community, meaning agents cannot interactively search or read Notion pages during a conversation.

Prior art

• PR #33645 proposed documenting an external PyPI package (langchain-notion-tools) but was closed because docs have moved to a separate repo. No native tool code exists in langchain_community.
• Several older closed issues requested Notion write/search capability but were never implemented.

This proposal fills that gap natively inside langchain_community.

Proposed solution

Add two tools to langchain_community/tools/notion/:

• NotionSearchTool — searches workspace via POST /v1/search
• NotionGetPageTool — fetches page blocks via GET /v1/blocks/{id}/children

Implementation plan

• Full BaseTool interface with Pydantic v2 validation
• SecretStr for API key handling (never logged or exposed)
• Google-style docstrings with Args, Returns, and Example sections
• Unit tests with fully mocked HTTP calls
• Proper registration in all init.py files

API reference

Notion public API (free, user-provided integration token) https://developers.notion.com/reference/intro

Happy to implement this if maintainers are open to it.

Use Case

Use cases

1. Meeting notes agent A team uses Notion as their knowledge base. An LLM agent can search Notion for past meeting notes relevant to a current topic and summarize them — without any manual copy-pasting.

2. Research assistant A user asks an agent "find everything we have on competitor X." The agent searches Notion, retrieves relevant pages, and compiles a structured summary on the fly.

3. Onboarding bot A new employee asks a Slack bot questions. The bot uses NotionSearchTool to find relevant internal docs and answers grounded in the company's actual Notion workspace.

4. Personal productivity agent A user says "find my Q3 goals page and tell me what I haven't completed." The agent searches Notion, reads the page content, and cross-references it with other context.

5. RAG over live Notion data Instead of a static document dump, an agent dynamically fetches the latest version of a Notion page at query time — always working with up-to-date content rather than a stale snapshot.

Proposed Solution

Proposed solution

Add two native tools to langchain_community/tools/notion/:

NotionSearchTool

• Calls POST https://api.notion.com/v1/search
• Accepts a natural language query string
• Returns matching page titles and their IDs
• Useful for agents that need to locate relevant pages dynamically

NotionGetPageTool

• Calls GET https://api.notion.com/v1/blocks/{page_id}/children
• Accepts a Notion page ID (returned by NotionSearchTool)
• Returns the readable text content of that page
• Useful for agents that need to read and reason over page content

These two tools work together naturally in an agent loop:

1. Agent calls NotionSearchTool("Q3 OKRs") → gets page ID
2. Agent calls NotionGetPageTool(page_id) → reads the content
3. Agent reasons over the content and responds to the user

What makes this different from the existing NotionDBLoader:

• NotionDBLoader is a one-time document loader for indexing
• These are agent tools — callable mid-conversation, on demand
• No need to pre-load or re-index when data changes in Notion

Alternatives Considered

No response

Additional Context

No response