# Vaner Documentation

> Vaner is a predictive context engine. It ponders your codebase in the background —
> exploring scenarios, pre-computing context, ranking what matters — so when you ask,
> the right evidence is already at the front of the queue.

## AI install & operation prompts

- [Install prompt](https://vaner.ai/prompts/install.md): install, wire MCP, pick backend, verify.
- [Upgrade prompt](https://vaner.ai/prompts/upgrade.md): upgrade in place, restart, verify.
- [Backend swap prompt](https://vaner.ai/prompts/backend-swap.md): switch model backend, verify.
- [Debug prompt](https://vaner.ai/prompts/debug.md): diagnose MCP not showing, silent ponder loop, stale context.

## Full documentation content

- [llms-full.txt](https://docs.vaner.ai/llms-full.txt): all documentation pages in a single file.

---

# Vaner

- [Introduction](/): Vaner is a local-first preparation engine that turns idle compute into evidence-backed Prepared Work for your AI client.
- [Getting Started](/getting-started): Install Vaner Desktop, connect your AI client, and start using Prepared Work.
- [Prepared Work](/prepared-work): What Vaner produces and how to inspect it.
- [Auto Focus](/auto-focus): How Vaner decides whether to work, where to work, and why.
- Integrations
  - [Integrations](/integrations): MCP-first integration paths for IDEs, runtimes, and optional compatibility layers.
  - [Connect your AI client](/integrations/connect-your-client): One command per AI client (Claude Code, Cursor, Codex CLI, Cline, Continue, Zed, Windsurf, Roo, VS Code, Claude Desktop). Vaner installs every applicable layer.
  - [MCP Mode](/integrations/mcp): Conceptual overview of Vaner's MCP integration and current tool surface.
  - [Context-only HTTP Mode](/integrations/context-only): Retrieve context packages over HTTP without proxying model requests.
  - [Python SDK Mode](/integrations/python-sdk): Use Vaner directly from Python applications and automation scripts.
  - [Proxy Mode](/integrations/proxy): Use Vaner as an OpenAI-compatible endpoint with context injection.
  - [Docker](/integrations/docker): Run Vaner daemon and API services with Docker Compose.

- **Reference**
- [CLI Reference](/cli): Core Vaner CLI commands for setup, Prepared Work, MCP, cockpit, and diagnostics.
- [Configuration](/configuration): Configure MCP, exploration, privacy boundaries, and optional gateway behavior.
- [Backends and Multi-Agent Setup](/backends): Pick the model endpoint Vaner uses for MCP-first pondering, with optional advanced gateway overrides.
- [Supported hardware](/supported-hardware): Vaner runs everywhere Ollama runs. Practical guide to which GPU, CPU, and platform combinations work, and which model size to expect on each tier.
- [Architecture](/architecture): How Vaner runs beside your AI client.
- [Security and Privacy](/security): Vaner is designed for local-first operation with explicit scope control.
- [Changelog](/changelog): Notable changes in Vaner documentation and releases.

- **More**
- [Troubleshooting](/troubleshooting): Quick fixes for common first-run issues with Vaner, MCP clients, and Ollama.
- [Examples](/examples): Ready-made examples for common Vaner integration patterns.