MCP-Native · Self-hosted · CI-synced

See how your system
fits together.

Connect your services, APIs, data, and diagrams into one view. Shared context for engineers and AI agents.

No credit card required

UIGraph — system mapsynced 4s ago
/checkoutPay nowPOST /v1/checkoutapi-storefront · OpenAPIGET /v1/pricingapi-storefront · OpenAPIcheckout-serviceowner · commercePRODpricing-serviceowner · platformPRODpayments-serviceowner · fintechPRODordersPostgreSQLledgerDynamoDB
checkout-service↑ 2 upstream · ↓ 1 downstream
acme/checkout-serviceOpen#1842
uigraph-botbotcommented just now

feat: add promo_code to checkout

UIGraph Impact Analysis CI Check: Passed

Changed
POST /v1/checkout
Downstream
2 services touched (pricing-service, orders-db)

Architecture Drift Detected

Code changes caused 2 artifacts to drift out of sync:

  • openapi.json (1 new parameter added)
  • checkout-sequence.mermaid (promo_code step added)
Create Sync PRView Drift Details

Synced via UIGraph MCP

Engineers — click any node to see what it calls, and what calls it

CI — impact, drift, and auto-sync posted on the PR

A system is more than its parts

The whole picture.
And the detail behind it.

Look closely at any individual component without ever losing sight of the bigger picture.

synced 4s ago

Explorer

  • .uigraph/
  • apis/
  • diagrams/
  • schemas/
  • dependencies/
.uigraph/diagrams/checkout-sequence.mermaid
Generated into .uigraph/ on every CI sync

Engineers browse · Agents query the same artifacts via MCP

Maintain it where you work

Your context.
Kept in sync.

Reference your existing artifacts in a UIGraph YAML file. Sync them using the CLI, or add the command to your CI/CD pipeline to update UIGraph on every deployment.

.uigraph.yaml
version: 1
service:
  name: checkout-service
  category: Backend
  repository:
    provider: github
    url: https://github.com/acme/checkout-service
  ownership:
    team: commerce

apis:
  - name: checkout-api
    type: openapi
    path: ./api/openapi.yaml

architectureDiagrams:
  - name: Checkout flow
    path: ./docs/checkout-flow.mmd
mcp — zsh
$ npx @uigraph/mcp
✓ connected to uigraph-mcp · 24 tools

  list_services     get_service
  get_api_spec      get_db_schema
  get_diagram       get_map
  search_docs       list_teams
  get_test_run
  + 15 more tools

# Claude / Cursor now answer from your real system

The same context for AI

Give your AI
the bigger picture.

Connect your coding agents to UIGraph through MCP so they can understand your system beyond the files they’re working on. Bring the same context into Slack, where your team can ask how something works and explore the artifacts behind the answer.

MCP serverClaude CodeSlack agent

Self hostable

Your system.
Your infrastructure.

Run UIGraph in your own infrastructure and keep your engineering context under your control. Manage team access with SSO and roles, and choose which AI providers you connect.

terminal
$ git clone https://github.com/uigraph-oss/uigraph-deploy.git
$ cd uigraph-deploy
$ make docker-up

✓ api        → http://localhost:8080
✓ ui         → http://localhost:3000
✓ postgres, redis, minio bundled

# First service map in under 5 minutes

See your own system
in UIGraph.

Bring in your engineering context and start exploring how it connects.

Free tier · Team & Enterprise plans · Commercial license for production use