Building an AI Assistant in C# with Cross‑Language Plugins

Description and use

Building a C# AI assistant using WeavePort to integrate existing Python/TypeScript plugins without rewriting, maintaining workflow control and security.

Problem

Developing a C# AI assistant while reusing existing Python/TypeScript libraries without rewriting them, while maintaining control over workflow, security, and lifecycle.

Outcome

WeavePort enables seamless integration of cross‑language plugins, preserving application control, security, and lifecycle management while avoiding code duplication.

Workflow

• 1. Identify a capability to replace (e.g., proposal scoring).
• 2. Create a plugin in the target language (Python) exposing a function (strategy.evaluate).
• 3. Register the plugin with WeavePort and bind it to a tenant/knowledge profile.
• 4. In C# host, instantiate LocalPluginClient and call CallAsync with request data.
• 5. Plugin receives request, calls host APIs (e.g., knowledge.read) via context.call_host.
• 6. Plugin computes scores and returns a typed result.
• 7. Host validates the result and commits it to application state.
• 8. WeavePort manages worker lifecycle, timeouts, and cleanup automatically.

Evidence

WeavePort is an open‑source plugin runtime for .NET.

Source

Python plugins run in Python processes. TypeScript plugins run through Node.

The host decides which records that grant permits.