kogakure/website-astro-stefanimhoff.de
1
0
Fork 0
mirror of https://github.com/kogakure/website-astro-stefanimhoff.de.git synced 2026-02-03 20:15:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: add section about using the MCP protocol with Readwise MCP in Raycast

Browse Source
This commit is contained in:
Stefan Imhoff
2025-05-29 13:45:53 +02:00
parent 5822cd77b9
commit c9b6d40e70
3 changed files with 6 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

BIN
public/assets/images/posts/readwise-raycast-mcp.jpg Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 1.1 MiB

BIN
public/assets/images/posts/readwise-raycast-mcp.webp Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 300 KiB

6
src/content/journal/2025/note-taking-obsidian-readwise-ai.mdx
View File

@@ -372,6 +372,12 @@ Readwise recently added the feature to [talk to your highlights and notes with A
In the future, I’ll likely use this feature every day to get answers for my notes. The good thing is that, in contrast to regular AI, which uses a web search to find the first few articles for a question and uses those as context, in Readwise I know that all the highlights I marked before as important and worth remembering make the answer much more valuable. But with all content, you get out what you feed it. If you read and highlight shallow articles or view things only from one side, your AI answers will be shallow and biased. In the future, I’ll likely use this feature every day to get answers for my notes. The good thing is that, in contrast to regular AI, which uses a web search to find the first few articles for a question and uses those as context, in Readwise I know that all the highlights I marked before as important and worth remembering make the answer much more valuable. But with all content, you get out what you feed it. If you read and highlight shallow articles or view things only from one side, your AI answers will be shallow and biased.
With the introduction of the [Model Context Protocol](https://modelcontextprotocol.io/) (MCP), an open standard that facilitates seamless integration between large language models (LLMs) and external data sources and tools, it is now possible to use external tools to interact with my highlights. Last month, Readwise released an [MCP](https://www.npmjs.com/package/@readwise/readwise-mcp), which I used to connect [Raycast](/raycast/) AI with their new [MCP feature](https://manual.raycast.com/model-context-protocol), allowing direct interaction with my highlights.
<Figure caption="Raycast AI with Readwise MCP">
<Image src="/assets/images/posts/readwise-raycast-mcp.webp" alt="Raycast AI with Readwise MCP" />
</Figure>
### Using AI for Research ### Using AI for Research
There are many people who think AI is bad for writing because it might replace the writer. However, AI is a great tool to accelerate the writing process and improve the understanding of the notes. I don’t see an issue with using AI to help me cut down on the time this work takes. I always read, check, and understand the output of AI and correct it when it produces incorrect output. There are many people who think AI is bad for writing because it might replace the writer. However, AI is a great tool to accelerate the writing process and improve the understanding of the notes. I don’t see an issue with using AI to help me cut down on the time this work takes. I always read, check, and understand the output of AI and correct it when it produces incorrect output.