The codegen.py module#
Summary#
Default pipeline: delegate to the backend. |
Description#
Extensible code-generation pipeline.
In Phase 1, the pipeline forwards (prompt, session_id) to the backend codegen method, which usually calls the existing Fluids One /api/chat/propose_code endpoint.
In Phase 3, the LLM orchestration loop moves into this module so the MCP server itself drives the generation by calling backend tools directly for named objects, state, and allowed values. Until then, this file gives every leaf a single, consistent place to evolve the pipeline.
Module detail#
- codegen.logger#