Spaces:
Running
Running
| # Qurio Python Migration Plan (Jan 22, 2026) | |
| ## Goal | |
| Migrate remaining Node.js LangChain endpoints to the Python (FastAPI + Agno) backend and align tool usage with Agno toolkits. | |
| ## Scope | |
| - Daily tip | |
| - Title | |
| - Title + space | |
| - Title + space + agent | |
| - Agent selection for auto mode | |
| - Related questions | |
| - Research plan (sync + SSE) | |
| - Deep research (SSE) | |
| - MCP tools endpoints | |
| ## Steps | |
| 1. Map existing Node endpoints/services to Python routes and services. | |
| 2. Add shared LLM utilities for non-streaming calls and JSON parsing. | |
| 3. Port prompts/logic for each endpoint, using Agno adapter for model calls. | |
| 4. Implement deep research orchestration with step execution + final report streaming. | |
| 5. Add MCP tool manager and endpoints (placeholder until a Python MCP client is added). | |
| 6. Wire routes into FastAPI and verify imports. | |
| ## Status | |
| - [x] Endpoint mapping | |
| - [x] Shared LLM utilities | |
| - [x] Daily tip / Title / Title + space / Title + space + agent / Agent auto / Related questions | |
| - [x] Research plan (sync + stream) | |
| - [x] Deep research stream | |
| - [x] MCP routes (stub until MCP client available) | |
| ## Notes | |
| - MCP routes are live but will return an error until a Python MCP client library is installed and wired in. | |
| - Tavily academic search is implemented as a local tool; general web search uses Agno's Tavily toolkit. | |