# Bonsai Docs

> Documentation for Bonsai, a standalone local research cache CLI for AI agents.

Bonsai captures documentation pages as reusable, source-cited Markdown artifacts with freshness metadata, deterministic JSON output, and compressed variants for context budgeting.

## Start Here

- [Bonsai — Grounded research for AI agents](https://bonsai.rhino-inquisitor.com/): Bonsai often costs more tokens per research session than native web search, but scores higher on official-source grounding and leaves captured pages on disk. Benchmarked in Codex, Cursor, Antigravity, and Claude Code.

## Guide

- [Introduction to Bonsai](https://bonsai.rhino-inquisitor.com/guide/introduction): When native agent web search is sufficient, and when you need Bonsai: deterministic capture, site modules for enterprise docs, and a durable local research cache.
- [Getting Started with Bonsai](https://bonsai.rhino-inquisitor.com/guide/getting-started): Run Bonsai from an empty cache, fetch a documentation page, and inspect the stored artifact from the command line.

## Concepts

- [Caching and Freshness](https://bonsai.rhino-inquisitor.com/concepts/caching-and-freshness): Understand Bonsai cache keys, freshness tiers, stale revalidation, TTLs, and the storage rules that keep docs research reusable.
- [Compression and Token Budgeting](https://bonsai.rhino-inquisitor.com/concepts/compression): See how Bonsai creates compressed and detailed Markdown variants so agents can fit complete technical docs into constrained context windows.
- [Storage Modes](https://bonsai.rhino-inquisitor.com/concepts/storage-modes): Compare Bonsai global and project storage modes, including lookup fallback, committed team caches, and secret-safe routing.
- [Bonsai vs Agent Memory](https://bonsai.rhino-inquisitor.com/concepts/vs-memory): Compare Bonsai with agent memory and learn why source-cited, freshness-aware documentation cache artifacts solve a different problem.

## How-to Guides

- [Share a Bonsai Cache with Your Team](https://bonsai.rhino-inquisitor.com/how-to/share-cache-with-your-team): Configure project storage, commit shared research artifacts, and let teammates or CI reuse the same captured documentation cache.
- [Cache Pages Bonsai Can't Fetch](https://bonsai.rhino-inquisitor.com/how-to/cache-pages-you-cant-fetch): Import authenticated, blocked, localhost, or JavaScript-rendered pages into Bonsai when automatic static fetching is not enough.
- [Import Synthesized Research](https://bonsai.rhino-inquisitor.com/how-to/importing-synthesis): Import manually synthesized notes, multi-source research, and durable metadata into Bonsai while preserving searchable cache artifacts.
- [Agent Integration](https://bonsai.rhino-inquisitor.com/how-to/agent-integration): Drive Bonsai from coding agents with cache-first lookup, deterministic JSON output, status checks, and source-cited documentation retrieval.
- [Install the Bonsai Agent Kit](https://bonsai.rhino-inquisitor.com/how-to/agent-kit): Install Bonsai agent instructions, skills, and subagents so AI coding tools research official docs through a reusable local cache.

## Reference

- [Bonsai Command Reference](https://bonsai.rhino-inquisitor.com/reference/commands): Reference every Bonsai command, positional argument, flag, JSON output envelope, and configuration command for agent-safe CLI use.
- [Bonsai Configuration](https://bonsai.rhino-inquisitor.com/reference/configuration): Configure Bonsai storage and summary behavior with flags, environment variables, project config, user config, and built-in defaults.
- [Cache Protocol Specification](https://bonsai.rhino-inquisitor.com/reference/cache-protocol): Review the Bonsai cache file format, YAML frontmatter fields, freshness lifecycle, revalidation behavior, and artifact metadata schema.
- [Site Modules Reference](https://bonsai.rhino-inquisitor.com/reference/site-modules): Learn how Bonsai site modules match documentation hosts, discover source URLs, extract page content, and preserve provenance metadata.
- [Bonsai Glossary](https://bonsai.rhino-inquisitor.com/reference/glossary): Definitions for Bonsai cache artifacts, freshness states, capture methods, storage modes, and documentation research workflow terms.

## Benchmark

- [Agent Research: Bonsai vs Native Web Search](https://bonsai.rhino-inquisitor.com/examples/agent-research-comparison): Codex, Cursor, Antigravity, and Claude Code session comparison of Bonsai-backed research versus native web search on TanStack Query, React Server Components, and Salesforce B2C Commerce prompts.

## More

- [Troubleshooting and Limits](https://bonsai.rhino-inquisitor.com/troubleshooting): Diagnose Bonsai fetch failures, network limits, SSRF protection, redirects, oversized responses, browser rendering, and cache issues.