# Belfort

> Documentation for Belfort's compilation-as-a-service platform for encrypted inference.

Every page of the documentation is also served as Markdown: add `.md` to its URL (the introduction is `/docs/index.md`). [/llms-full.txt](/llms-full.txt) holds all of them in one file.

## Getting started

- [Quickstart](/docs/getting-started/quickstart.md): From a PyTorch model to an encrypted inference
- [Introduction](/docs/index.md): Documentation for Belfort's compilation-as-a-service platform for encrypted inference.
- [Installation](/docs/getting-started/installation.md): Install belfort-ml and belfort-ml-client, and get an API key.
- [Examples](/docs/getting-started/examples.md): The example models, and how to compile one.
- [Coding agents](/docs/getting-started/coding-agents.md): Teach Claude Code or another coding agent to build on Belfort with the Belfort skill.
- [Troubleshooting](/docs/getting-started/troubleshooting.md): Common errors encountered while compiling and running an encrypted model

## Concepts

- [Model providers and end clients](/docs/concepts/providers-and-end-clients.md): Who does what, and what never leaves the end client.
- [Compilation pipeline](/docs/concepts/compilation-pipeline.md): From a PyTorch model to a GPU-ready FHE circuit.
- [Keys and encryption](/docs/concepts/keys-and-encryption.md): Key sets, secret keys and CKKS ciphertexts.
- [Deploys and credits](/docs/concepts/deploys-and-credits.md): GPU deploys, lifetimes and how credits are spent.

## SDK

- [Provider](/docs/sdk/provider.md): Configuring and using `Provider`.
- [Errors and retries](/docs/sdk/errors-and-retries.md): The error hierarchy and retries.
- [The Belfort ML client](/docs/sdk/belfort-ml-client.md): Running the Belfort ML client from belfort-ml-client on a client token.

## Reference

- [API reference](/api/docs)

## Agent skills

- [belfort](/skills/belfort/SKILL.md): a skill that teaches a coding agent to use Belfort; [/skills/belfort.zip](/skills/belfort.zip) for apps that upload one