trustfoundry-ai turns Mike's chat into a socket for outside tools
The team built a plug-in system so third-party legal tools can drop into the assistant chat without anyone rewiring the core.
Until now, adding a tool to Mike's chat assistant meant editing shared code by hand. This work introduces a plug-in interface: each provider ships as a self-contained module carrying its own tools, labels, and logic, so new ones can be added without touching everyone else's code. To prove it out, the team wired up their own TrustFoundry service as the first integration, adding three tools that appear in chat only when it's switched on:
- Legal search - run legal-research queries without leaving the chat.
- Citation validation - check whether a cited authority actually exists.
- Usage tracking - see how much of the service you're consuming.
Two smaller touches round it out: the chat now shows a friendly status like "Running TrustFoundry Legal Research..." instead of a raw tool name, and unrecognized tool calls surface a clear error instead of failing silently. One caveat: this targets tools secured by a single shared key, so providers that need per-user logins aren't covered yet.
Spotted something wrong? Or know the PR text has fresher detail than the writeup above?