Website · Docs · Releases · Issues
接入你的智能体
從發布者安裝。不是 PlugMCP 端點。
遠端 URL 還是本機指令? 兩種安裝方式的差異
Enterprise knowledge is trapped in silos. Digital tools (Slack, GitHub, Notion, Gmail) don't talk to physical systems (IoT sensors, industrial protocols, camera feeds). OpenBeam unifies both into one searchable, agent-ready layer. One query across sensors and SaaS. Answers in 200ms. Runs on your servers.
Glean proved the digital half at a $7.2B valuation. Nobody has built the physical half. We're building both.
Digital Knowledge
Communication
IoT Platforms
Industrial Protocols
Plus 60+ more across CRM (Salesforce, HubSpot, Pipedrive, Zendesk), Code (GitHub, GitLab, Bitbucket, Jenkins), Storage (Google Drive, Dropbox, Box, S3, SharePoint), Project Management (Linear, Jira, Asana, ClickUp, Monday), Design (Figma, Canva, Miro), HR (Workday, BambooHR, Greenhouse), and more. Full connector list →
Cross-platform Go binary, signed with cosign, with SLSA Level 3 build provenance.
# macOS — Homebrew
brew install kuluruvineeth/tap/openbeam
# Linux / macOS — curl
curl -fsSL https://openbeam.work/install.sh | bash
# Windows — Scoop
scoop bucket add openbeam https://github.com/kuluruvineeth/scoop-bucket
scoop install openbeam
# Docker
docker run --rm ghcr.io/kuluruvineeth/openbeam-cli:latest version
Or download .deb, .rpm, .apk, Arch, or platform binaries from the latest release.
openbeam auth login --api-key op_live_xxx
openbeam search query "kubernetes upgrade runbook"
openbeam upgrade # self-update
openbeam doctor # diagnose install + auth + connectivity
openbeam computer run knowledge-digest --wait
Six pre-built agents that run on a schedule, ask for approval before any write, and remember context across runs.
Each agent is durable via Temporal, multi-surface (CLI / Dashboard / Chat / API / MCP), and approval-gated for any third-party write action. Custom agents are defined by a prompt and a tool list.
openbeam computer enable knowledge-digest --schedule "0 9 * * MON"
openbeam computer run knowledge-digest --wait
openbeam computer approve <run-id> --actions "archive,notify"
git clone https://github.com/kuluruvineeth/openbeam.git
cd openbeam
make setup # Install deps, start infra, setup database
make apps # Start all apps with hot reload
Open http://localhost:3001 and you're in.
docker compose -f docker-compose.infra.yml -f docker-compose.yml up -d
See the Self-Hosting Guide for configuration, TLS, monitoring, and scaling.
make setup # First-time setup (deps, infra, database)
make dev # Start infrastructure services
make apps # Start all apps with hot reload
make status # Show service health
make down # Stop infrastructure
bun run check # Lint & format (Biome)
bun run build # Build all packages
bun test # Run tests
Enterprise search touches your most sensitive data — every message, document, sensor reading, and credential. You should be able to read every line of code that processes it.
AGPL-3.0 licensed. Self-host it, audit it, extend it. No vendor lock-in. No data leaving your network.
We welcome contributions. See CONTRIBUTING.md for guidelines.
git clone https://github.com/YOUR_USERNAME/openbeam.git
cd openbeam
make setup && make apps
Connectors are the fastest path to a merged PR — each takes about 2-3 hours. Check the good first issue label to get started.
Built by @kuluruvineeth
Website · Docs · Pitch Deck · Book a Demo
Open-source Glean for SaaS and the physical world. 87 connectors, hybrid search, autonomous agents, MCP server. Self-hosted, AGPL.
工具
尚無工具清單。遠端伺服器會在首次健康檢查後取得工具清單。
常見問題
在本頁複製適用於 Cursor、Claude Desktop、Claude Code 或 VS Code 的 stdio 設定。它使用發布者的安裝方式,而不是 PlugMCP 代理伺服器。
這筆記錄未列出必要的環境變數。
Openbeam 透過 stdio 在本機執行。
上次檢查時間:2026-10-28。 Open-source Glean for SaaS and the physical world. 87 connectors, hybrid search, autonomous agents, MCP server. Self-hosted, AGPL.