The AI coding IDE · v1.35.40 · MIT

Rabbit is your AI coding agent for building ambitious software.

Built for developers who ship with local models, hybrid cloud routing, and an MCP token economy that slashes context spend. Local-first by default — no editor telemetry, no cloud code index, account optional. No refusals. No moralizing. Just code.

Download for macOS See it in action

macOS·Linux·Windows·arm64 + x64

rabbit · agent

~ / project

$rabbit agent run --task 'add jwt auth + tests'
▸ctx_tree . — 12k → 1.4k tokens (compressed)
▸ctx_read src/auth.ts (signatures) — 320 tokens
▸graph.related → 4 files (pruned)
▸routing → local: llama-3.1:70b (planning)
▸routing → cloud: claude-sonnet-4 (codegen)
✓applied 3 edits · added 14 tests · all green
$

Tokens saved · this session

vs. running the same agent loop on Cursor with the same model.

ctx_read cache hits412

signatures mode reads68

cloud calls avoided231

$ saved (estimate)$14.20

Trusted by developers who build with every major model and open standards

VS Code OSS

MCP

Anthropic

OpenAI

Ollama

Llama

Qwen

DeepSeek

Mistral

Tree-sitter

LSP

Tauri-ready

Token economy

Stop renting tokens you already paid for.

Rabbit's MCP layer (stoken) caches reads, compresses output, and prunes context with a project-wide dependency graph. The result: the same task, a fraction of the tokens.

ctx_read cache

cache hits

Re-reading the same file costs ~13 tokens instead of 500+. Cache survives sessions.

signatures / map modes

smaller reads

Read only the AST skeleton when you don't need the full file. Imports, exports, types.

ctx_shell compression

compressed

Strips ANSI, deduplicates noise, keeps essentials. 90+ commands recognized.

graph-pruned context

pruned

Dependency graph trims irrelevant files before they touch the prompt.

Illustrative example

"Refactor auth module + write tests"

Context sent

less

Typical AI IDE140k tokens

Rabbit (stoken)35k tokens

$ ctx_read src/auth.ts → full | 1.2k tokens

$ ctx_read src/auth.ts → cached | 13 tokens

$ ctx_read src/auth.ts mode=signatures → 87 tokens

Local models

Run any model. On your machine.

Rabbit ships with first-class support for local inference. Plug in Ollama or any OpenAI-compatible endpoint and the agent just works — with full tool calling, streaming, and project context.

Llama 3.1

Qwen 2.5 Coder

Alibaba

1.5B / 7B / 32B

ollama pull qwen-2.5-coder

DeepSeek-Coder V2

DeepSeek

16B / 236B

ollama pull deepseek-coder-v2

Mistral / Codestral

Mistral

7B / 22B

ollama pull mistral-/-codestral

Phi-3.5

Microsoft

3.8B / 14B

ollama pull phi-3.5

Gemma 2

Google

2B / 9B / 27B

ollama pull gemma-2

Air-gapped by default

Your code never leaves the machine when you use a local model. No editor telemetry, no cloud code index.

Zero rate limits

Burn through 10M tokens an hour if your GPU lets you. No throttling, no quotas.

Bring your own runtime

Ollama, llama.cpp, vLLM, LM Studio, MLX — anything with an OpenAI-compatible endpoint.

Apple Silicon tuned

MLX kernels and Metal acceleration out of the box on M-series machines.

Hybrid routing

Cheap locally. Smart in the cloud.

Rabbit routes each agent step to the right model. Plan and search locally for free. Send only the hard codegen step to Claude or GPT. Your wallet feels the difference immediately.

Per-step routing rules in YAML, hot-reloadable.
Fallback chains — local → mid-tier cloud → frontier.
Per-project budget caps with hard cutoffs.
Outsource entire tasks via the `rabbit-cli` headless runner.

routing.yml

agent:
  plan:       ollama/llama3.1:70b   # local · free
  search:     ollama/qwen2.5-coder  # local · free
  codegen:    anthropic/sonnet-4    # cloud · paid
  review:     openai/gpt-4.1-mini   # cloud · cheap
  fallback:
    - ollama/qwen2.5-coder:32b
    - openai/gpt-4.1-mini
  budget:
    daily_usd:   5.00
    hard_cutoff: true

Local plan

Cloud codegen

Local review

No moral filters

The model writes code. Not lectures.

Rabbit is built for adults shipping production software. The default agent prompt strips refusals, removes unsolicited warnings, and disables the apology loop. You wrote the prompt — the model writes the code.

No "I'm sorry, but I can't" boilerplate.
No unsolicited security/ethics monologues.
No silent edits "for your own good".
Pair with any local model for fully uncensored loops.

I cannot help with that request.stripped

As an AI assistant, I must decline...stripped

I'm not able to write code that...stripped

This could be used for harmful purposes...stripped

Have you considered the ethical implications?stripped

Here is the diff. Tests are passing.shipped

Privacy · local-first

The IDE that doesn't watch you.

Rabbit is local-first by design. No editor telemetry, no cloud code index, no prompt logging. Cloud features are opt-in and only touch what they need.

No telemetry in the IDE

No analytics SDK, no crash pings, no usage events from the editor. Usage is recorded only when you opt into Rabbit Cloud routing — and only to bill it.

No account required

Download. Run. Code. An account is only needed for the optional cloud proxy, subscriptions, and auto-updates.

Fully offline-capable

Pair with a local model and pull the network cable. Agent, autocomplete, search, MCP, refactors keep working.

Your keys, your machine

Provider API keys are stored locally in your settings and sent only to the provider you chose. Never synced to our servers, never logged.

Your code stays local by default

Indexing, memories, sessions live in ~/.rabbit-ide/. Code is sent only to the model endpoint you configured — local or cloud.

No vendor lock-in

Open formats for memories, rules, sessions. Walk away whenever you want — everything is plain JSON on your disk.

Our pledge

Collect the minimum. Keep it honest.

We can't sell what we don't collect. If you never sign in, we hold nothing about you. If you do, we store only your account, balance and usage — nothing from your codebase.

# default privacy posture — every install

[telemetry]
enabled       = false   # no editor analytics
crash_reports = false   # not collected

[cloud]
account_required = false  # optional, for proxy/updates
prompt_logging   = false  # prompts are not stored
code_upload      = false  # no cloud code index

[storage]
location = "~/.rabbit-ide/"
sync     = "off"          # opt-in only

Agent · MCP native

An agent with 58 native tools. Plus every MCP server you bring.

The agent loop ships with the stoken MCP server preconfigured. Drop any other MCP server into ~/.rabbit-ide/mcp-servers.json and its tools appear automatically in the prompt.

Drag Rabbit IDE - tools follow, tap a tool to zoom, tap a group to isolate

58 native MCP tools orbit Rabbit IDE — drag the center, click any spoke to explore.

~/.rabbit-ide/mcp-servers.json

{
  "stoken":   { "command": "stoken",   "args": ["serve"] },
  "postgres": { "command": "mcp-pg",   "args": ["--url", "$DATABASE_URL"] },
  "puppeteer":{ "command": "mcp-pptr", "args": [] },
  "figma":    { "command": "mcp-figma","env": { "FIGMA_TOKEN": "$FIGMA_TOKEN" } }
}

Compatibility

Drop in. Nothing to relearn.

Rabbit is a VS Code OSS fork. Your extensions, keybindings, settings, and workspace files all work. Cursor rules, Devin workflows, and AGENTS.md from Antigravity import automatically.

VS Code extensionsFull Marketplace + OpenVSX

.cursorrulesImported on first open

.devin/ configWorkflows, hooks, permissions

AGENTS.mdCursor, Devin, Antigravity rules

MCP serversPer-workspace and global

LSP / DAPInherited from VS Code OSS

Settings syncSelf-hosted, optional

Telemetry—No editor analytics SDK

rabbit-cli

Same agent. In your terminal.

rabbit-cli is the IDE agent without the editor — headless, scriptable, CI-friendly. Same routing config, same MCP tools, same prompt.

use

CI guards · code review bots

use

Migrations · refactors at scale

use

Outsourced tasks · overnight runs

use

SSH into a remote box, run it there

~ / acme-monorepozsh

$ npm i -g @rabbit/cli
$ rabbit login --provider ollama
$ rabbit run "bump every workspace to node 22 and fix
              every type error, then open a PR"

  ▸ scanning 142 packages
  ▸ planning: local · llama3.1:70b
  ▸ editing:  cloud · sonnet-4
  ▸ 38 files changed · 0 type errors · 312 tests green
  ✓ opened PR #4821

Pricing