
jongodb-labs/bedrock-now
Developer toolsЭта страница переведена автоматически. Технические детали смотрите в оригинале. Открыть оригинал
One-command Docker stack for AWS Bedrock: LiteLLM gateway, Open WebUI, shared S3 + MCP for Claude Code, per-key budgets and cost transparency with option OpenRouter integration
Обзор
bedrock-now is a private, single-user AI workbench for your own Mac or Linux machine: a chat app, Claude Code and a shared file store, with every model call billed to under your team's AWS access key. You get modern models without sending your work to a consumer AI service, and you see what every answer costs. Install it with one command; nothing listens beyond 127.0.0.1. - macOS or Linux, with (macOS) or (Linux) - git, curl, openssl (macOS's built-in bash 3.2 is fine) - An for an IAM user with Bedrock access (long-lived AKIA… key; see Asking your AWS admin for model access), or, on an EC2 VM, an instead (no keys stored; see Run it on a cloud VM) - Optional: (claude), for claude-bedrock The installer checks the prerequisites and says how to install anything missing, clones the latest release into ~/bedrock-now (BEDROCK_NOW_DIR=/path for another folder, BEDROCK_NOW_REF=main for the newest code) and starts setup.
README
Показан фрагмент оригинального README. Открыть оригинал.
DevOS ADE
DevOS ADE (formerly bedrock-now) is a private, single-user AI workbench for your own Mac or Linux machine: a chat
app, Claude Code and a shared file store. Models run on Amazon Bedrock in your own AWS account (an IAM user’s
access key, or an EC2 instance role) and are billed there; the only other places a call can go are ones you turn on
yourself — OpenRouter with your own key (billed by OpenRouter; some models are free), the Experiential Labs
gateway with your own key (its credits, or your own provider keys connected there) and speech-to-text through OpenAI or
Groq with your own key. The microphone in Chat uses your browser’s own speech recognition unless you choose
otherwise, and that sends what you say to the browser’s maker (Google, Microsoft or Apple; Voice).
You get modern models without sending your work to a consumer AI service, and you see what every answer costs.
Install it with one command; nothing listens beyond 127.0.0.1.
What you get
Everything below runs on your computer, behind one portal sign-in. The optional parts are off until you turn them on;
the Code page and its browser terminal are on by default (./stack terminal disable turns both off, ./stack code disable the Code page alone; either stays off across updates).
| Part | What it is | Default | On / off |
|---|---|---|---|
| Chat | Open WebUI with the Bedrock models your key can use (and OpenRouter’s, once you turn it on): the cost under every answer, documents and knowledge bases, images | on | — |
devos-claude |
Claude Code in your own terminal, billed to Bedrock (or to OpenRouter for its models), with a lean config folder and each session’s cost; any chat model of the catalog (--models lists them and how each works, --model picks one) |
on (needs claude) |
setup offers to put it on your PATH |
| Home, Admin, Files | health, spend and budgets; LiteLLM’s UI; the shared file store (also an MCP server for Claude Code) | on | — |
| The API for your tools | the same models for your own scripts and apps: OpenAI- and Anthropic-compatible, on 127.0.0.1, a key and budget per tool (Admin › Virtual Keys) |
on | — |
| Activity & search | a bell for budget alerts, updates, Code sessions that need you and jobs; Ctrl/⌘+K finds pages, settings, Code sessions and workspace files, Chat conversations, the walkthrough and these docs | on | — |
| Settings | every setting of .env, shown in the portal |
on (read-only until the update helper is installed) | ./stack helper install makes it editable |
Установка
This server does not publish a one-line install command.
Open the repository installation guide Новичок в MCP? Начните с руководства на mcpskill.org →