Declare integrations and custom schemas in a single YAML file. Get a braided API. Ships as a single Go binary.
# braids.yaml — merge Stripe + Shopify customers version: "1" connectors: stripe: type: stripe auth: api_key: ${STRIPE_SECRET_KEY} shopify: type: shopify auth: api_key: ${SHOPIFY_API_KEY} store: ${SHOPIFY_STORE} endpoints: /customers: merge: on: email strategy: deep_merge sources: - connector: stripe resource: customers fields: id: stripe_id email: email name: full_name - connector: shopify resource: customers fields: id: shopify_id email: email first_name + last_name: full_name
$ braids serve ■ Braids Gateway v0.1.0 ──────────────────────────────────── ✓ Config loaded braids.yaml ✓ Connector ready stripe ✓ Connector ready shopify ✓ Schema validated 1 endpoint, 2 sources ✓ Merge strategy deep_merge on "email" ■ Gateway listening on http://localhost:8080 Endpoints: GET /customers stripe + shopify Press Ctrl+C to stop
AI can write your Stripe integration in an afternoon. But who maintains it next quarter when the API changes, the second source gets added, and the schema drifts?
Every integration becomes a liability. More sources, more glue code, more things that break at 2am.
Declare your integrations in YAML. Braids handles the fetching, mapping, merging, caching, and error handling.
Update one field — not fifty files.
Ship fast with AI. Stay fast with config.
Everything lives in one YAML file. GitOps-native, diffable, reviewable. No scattered integration code.
You define the output shape. Braids maps upstream data into it. Your API, your contract.
Single Go binary. Deploy with Docker, Kubernetes, bare metal, or let us host it on Braids Cloud.
Auto-generate field mappings with your own LLM key. Anthropic, OpenAI, or fully offline via Ollama.
Share and install connectors, schemas, and mapping recipes. Build on what others have done.
Free forever. MIT licensed. Full transparency. No vendor lock-in, no surprises.