HLVM Docs
HLVM is a vendor-neutral AI runtime. Start with the smallest path that proves your setup, then use the workflow pages when you need a specific surface.
Reading Paths
| Goal | Read |
|---|---|
| Build the pre-release source | Build |
| Prove the first AI run | First Run |
| Run one-shot agent tasks | Ask |
| Work interactively | REPL |
| Switch or download models | Models |
| Use named assistants | Agents |
| Connect external tools | MCP |
| Connect the browser bridge | Browser Bridge |
| Expose the local runtime server | Runtime Server |
| Diagnose a local build or runtime | Troubleshooting |
| Look up every command | CLI Reference |
The Short Path
mkdir hlvm-dev && cd hlvm-dev
git clone https://github.com/hlvm-dev/hql.git
git clone https://github.com/hlvm-dev/hlvm.git
cd hlvm
make build
./hlvm --version
./hlvm ask "summarize what HLVM is in one paragraph"
The repository is currently private and pre-release. No public binary installer is published yet; this path is for authorized contributors.
A fresh configuration starts at auto: HLVM chooses one usable route before
each turn. Runtime startup does not bootstrap local AI. Installed and
authenticated native engines and providers appear as eligible routes; local
setup remains an explicit option. User-facing clients talk to the runtime host
on 127.0.0.1:11435.
What To Read Next
- Contributors should read Guide after a local build.
- Shell-heavy users should read Ask and CLI Reference.
- Tool users should read MCP, Agents, and Browser Bridge.
- Architecture readers should use Agent Engine Federation for the current-versus-target ownership boundary.
- HQL users should switch to the HQL docs.