---

AI / Agent frameworks

These help build AI assistants, tools, and workflows.

Chainlit
The one you’re using now. It provides the chat UI and event system for AI apps.

LangChain
A toolkit for connecting LLMs, tools, memory, and retrieval pipelines.

LlamaIndex
Focused on retrieval-augmented generation (RAG) — connecting models to data sources.

You also used the OpenAI Agents SDK, which is more of an agent orchestration toolkit than a full framework.

---

Web frameworks

You’ve used these when building small apps or services.

Flask
A lightweight Python web framework you’ve used for tools like TransferDepot and other mini web services.

Backdrop CMS
A CMS framework you explored for structured websites and login-based pages.

MediaWiki
The platform behind your manual procedures system. It’s also a framework for structured content with extensions.

---

Infrastructure / platform frameworks

These structure how systems run rather than the code itself.

Podman
Your container platform for MediaWiki and other services.

Proxmox VE
Your VM orchestration layer in the Camelot environment.

---

Data / observability frameworks

Elasticsearch
The core of the ELK stack you’ve been exploring for logs and AI-assisted ops.

---

The pattern across all of them

Every framework basically gives you three things:

1. Structure
   Example: @cl.on_message in Chainlit or routes in Flask.
2. Plumbing
   Networking, UI, model calls, etc.
3. Conventions
   “Put this file here and the framework will wire it up.”

So your Chainlit app is basically:

Chainlit (framework)
     ↓
Runner (agent runtime)
     ↓
nutrition_agent (logic)
     ↓
calorie_lookup_tool (tool)
     ↓
ChromaDB (data)

---

The reason the nouns blur together

Because modern AI stacks layer multiple frameworks at once:

Chainlit
   + OpenAI Agents SDK
   + ChromaDB
   + Python tools
   + Render deployment

That’s why it can feel like you’re juggling ten new nouns.

---