Shogo AI
Shogo is an open-source platform for building AI agents that do real work: reading from your systems, taking actions, and running workflows end to end. TypeScript throughout. Self-host it or use Shogo Cloud.
Website · Launch Studio ·
接入你的智能体
發布者自己的指令。不是 PlugMCP 端點。
用我們的金鑰與日誌託管一份副本。
npx -y @shogo-ai/[email protected]Claude Code 沒有本機指令的安裝連結。複製下方指令並在終端機中執行。
claude mcp add shogo-ai-sdk -- npx -y @shogo-ai/sdk@2.5.8在終端機中執行。
{ "mcpServers": { "shogo-ai-sdk": { "command": "npx", "args": [ "-y", "@shogo-ai/[email protected]" ] } }}claude_desktop_config.json
遠端 URL 還是本機指令? 兩種安裝方式的差異
來自發布者
Shogo is an open-source platform for building AI agents that do real work: reading from your systems, taking actions, and running workflows end to end. TypeScript throughout. Self-host it or use Shogo Cloud.
Website · Launch Studio ·
Shogo Platform SDK - Zero-boilerplate auth and database for Shogo apps
Shogo combines a Hono API, Expo-based clients, agent runtimes, project runtimes, and a developer SDK into one platform for building and operating agentic products.
License: the SDK and client libraries you actually integrate
(@shogo-ai/*, plus the mobile/desktop clients) are MIT. The server
components behind Shogo Cloud (apps/api/, packages/agent-runtime/,
packages/shared-runtime/) are AGPL-3.0-or-later — a moat against hosted
resellers, not a restriction on your own use, modification, or
self-hosting. Full breakdown in Open Source Model
below and docs/LICENSING.md.
There's no one-command quickstart yet. Running Shogo locally means cloning the repo, installing with Bun, starting Postgres/Redis/MinIO in Docker, and running migrations — realistically a few minutes, not 60 seconds. The real steps are in Local development below.
Shogo uses a split-license model. AGPL-3.0-or-later guards the cloud-service surface a competitor would need to ship a hosted clone; everything else is MIT so adoption is friction-free.
apps/api/, packages/agent-runtime/,
packages/shared-runtime/@shogo-ai/* libraries, apps/mobile/, apps/desktop/,
packages/shared-app/, packages/shared-ui/, packages/ui-kit/,
packages/domain-stores/, templates/runtime-template/apps/docs/, docs/terraform/, k8s/, deploy-examples/,
.github/workflows/ (see INFRASTRUCTURE-LICENSE.md)See docs/LICENSING.md for the full strategy and
rationale, plus LICENSE, NOTICE, INFRASTRUCTURE-LICENSE.md, and
TRADEMARK.md.
License is shown inline so the AGPL/MIT boundary is visible at a glance.
Prerequisites:
bun install
cp .env.example .env.local
bun run docker:infra
bun run db:migrate:deploy
bun run build:packages
bun run dev:all
Open http://localhost:8081.
Shogo can be self-hosted locally or on your own infrastructure. For setup
details, required environment variables, storage, and deployment notes, see
docs/SELF_HOSTING.md.
Local desktop/offline usage is documented in apps/desktop/README.md.
Published to npm (MIT, lockstep release on the sdk-v* tag):
Old @shogo-ai/sdk/<subpath> imports keep working through deprecated
re-export shims; see packages/sdk/MIGRATION.md.
Workspace-only (AGPL):
Workspace-only (MIT):
CONTRIBUTING.md for contribution guidelinesCLA.md for contributor licensing termsSECURITY.md for responsible disclosureTRADEMARK.md for branding and name usagedocs/GETTING_STARTED.mddocs/ARCHITECTURE.mddocs/SELF_HOSTING.mdpackages/sdk/README.md{ "mcpServers": { "shogo-ai-sdk": { "command": "npx", "args": [ "-y", "@shogo-ai/[email protected]" ] } }}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。
@shogo Ai/sdk 透過 stdio 在本機執行。
上次檢查時間:2026-11-17。 Shogo Platform SDK - Zero-boilerplate auth and database for Shogo apps