Core user paths
- First chat
- Running the web UI and API
- Scheduling jobs
- Configuring agents, tools, skills, and memory
Documentation hub
A front door, not another markdown landfill.
term-llm has outgrown the giant README. This is the hub: clear paths, fewer dead ends, and room for the docs to become an actual product.
Choose your path
Start with intent, not file names
New here
Getting started
Install, configure, and get to a first successful run without archaeology.
Do a thing
Guides
Task-focused docs for providers, agents, memory, jobs, web, and Telegram.
Understand it
Architecture
How sessions, runtimes, tools, skills, memory, and routing actually fit together.
Need exact truth
Reference
CLI, config, env vars, API surface, and the sharp edges where hand-waving dies.
What belongs here
Documentation shape
- Quickstart flows that get from zero to useful
- How-to guides for concrete tasks
- Architecture pages that explain the model
- Reference pages that stop arguing and state facts
What does not
Things we are trying not to repeat
- One README containing tutorial, manifesto, and config dump in a trench coat
- GitHub directory spelunking mistaken for information architecture
- Pretty docs chrome hiding content that still has no classification
- Reference facts trapped inside prose paragraphs nobody can scan
First skeleton
Planned doc clusters
Operator and contributor paths
- Deploying runtimes
- Understanding session/state behavior
- Adding providers and tools
- Troubleshooting weird runtime failure modes