Why we built ContextPass

During the development of AI agents and MCP servers, we realized, that configuring and running MCP servers kind a sucks. Each MCP server requires access tokens from vairous cloud services or database usernames and passwords. All this credentials together with command and arguments for running servers should be added to each agent configuration. And, of course, we spend a lot of time to debugging server in MCP inspector.

We also found it difficult to understand how agents like Claude and Cursor interact with our MCPs, since existing observability tools can’t access these closed systems.

Overall experience with MCPs was not that great.

What we where missing

• One Place for All MCPs: Single inventory of servers — local or remote — that we could add to our AI agents without juggling command parameters and variables.
• Built-in Observability: Existing observability tools can’t connect to closed-source agents like Claude or Windsurf, making it hard to see how prompts, tools, and data are actually used.
• Simple Setup and Validation: we wanted to configure and test our MCP servers once, then reuse that setup across our agents and assistants. Running MCP inspector with different configurations was quite a hassle.
• Local-Only Analytics: we wanted to keep all usage data and analytics on our development machines.

What we have built

ContextPass is a lightweight desktop app that:

• Guides you through setup process and validates configurations.
• Shows all your MCP servers in one place (local and remote).
• Records agent interactions and usage data for troubleshooting and prompt optimization.
• Manages server configurations that connect to different data stores and APIs.
• Built-in MCP inspector for fast tool, prompt and resource testing and debugging.
• Keeps all collected data and analytics private and local.

Open Source

We have open sourced the code on GitHub under the MIT license. We wanted ContextPass to be able to reviewed, extended, and trusted by the community.

With ContextPass, we spend less time on server setup and more time on our AI projects. By sharing ContextPass, we hope it helps you do the same.