How to incorporate resources as context on the client side

[djwessel]

Pre-submission Checklist

• I have checked that this question would not be more appropriate as an issue in a specific repository
• I have searched existing discussions and documentation for answers

Question Category

• Protocol Specification
• SDK Usage
• Server Implementation
• General Implementation
• Documentation
• Other

Your Question

Hey there,

I'm looking into building a client to work with MCP servers, and I'd like to know how others solve the problem of including resources in the context.

The official spec leaves this question rather purposefully open-ended, which is understandable.

Resources in MCP are designed to be application-driven, with host applications determining how to incorporate context based on their needs.

For example, applications could:

• Expose resources through UI elements for explicit selection, in a tree or list view
• Allow the user to search through and filter available resources
• Implement automatic context inclusion, based on heuristics or the AI model’s selection

However, I was wondering what approaches others take to integrate the resources into an LLM's context, especially related to the third point of automatic context inclusion.

I tried looking into different open-source implementations that utilize MCP, including continue and cline and I found different approaches. Continue seems to rely on the user to identify when they want to inject a specific resource into the context, whereas cline seems to wrap resource injection as a tool call to allow the LLM to automatically identify which context should be included.

Does anyone else have ideas or experience in this topic regarding what works and what doesn't?