modelcontextprotocol
Include title of fetched page in content returned by the fetch server
Describe the bug
Fetching the URL https://pandas.pydata.org/docs/dev/whatsnew/v2.1.0.html does not include the the title line "What’s new in 2.1.0 (Aug 30, 2023)".
To Reproduce
Run the fetch tool with the https://pandas.pydata.org/docs/dev/whatsnew/v2.1.0.html URL.
Expected behavior "What’s new in 2.1.0 (Aug 30, 2023)" title line should be included, but it isn't.
Logs N/A
Additional context This was done as part of an Outlier project which implemented the tool. I have no access to their implementation of the MCP server but I don't think its a server configuration error.
@gebalassa could you clarify what you mean by "Outlier project which implemented the tool?" Do you mean that it was a different version of the fetch tool?
@gebalassa could you clarify what you mean by "Outlier project which implemented the tool?" Do you mean that it was a different version of the fetch tool?
Hello! Allow me to clarify.
If I recall correctly, the project intended to train AI to use multiple MCP tools in order to solve prompts. The tools were queried with calls in JSON format. One of these tools was a "fetch" tool, which, I understand, extracts text from a given URL.
In one attempt where I used it to obtain the contents from the Pandas URL mentioned above, the title line was not returned, though everything afterwards was.
In all honesty, due to the nature of the project, I cannot verify the used version, nor if the given "fetch" MCP tool is the exact same as the one in this repository, though the required parameters (a URL) were exactly the same, and this repository seems to be the most popular in this regard.
If the current version does effectively return the title of the above URL, please feel free to mark this as solved.
Thank you for your time.
Thanks for clarifying @gebalassa - I checked the implementation in this repo and it doesn't appear to include the page title. So I think this would be an enhancement.