CompanyManifestoCustomer storiesShip logCareersSandboxPricing
Log in
Platform · Developer

Extend ScendCore with your code.

REST API, signed webhooks, official SDKs, and custom skills. The same primitives we build agents on are open to you — build your own agents, tools, and workflows on top.

Most AI platforms ship a chat interface and call themselves extensible. ScendCore ships the primitives — every action an agent can take is an API call you can make too, every event is a webhook you can subscribe to, every skill is a tool manifest you can author. If we can build it, you can.

Powered by ScendCore Brain
Open primitives, same APIs
In this explainer
  1. 01REST API
  2. 02Webhooks
  3. 03SDKs
  4. 04Custom Skills
Phase 01

REST API · The execution surface

Every agent action, every contact, every conversation, every CRM write — accessible via a versioned, OpenAPI-described REST API. Token-scoped, rate-limited, fully audited.

Read access for reporting, write access for orchestration. Most teams wire up reporting first (Looker / Tableau pulls), then graduate to writes (trigger campaigns from your own systems, push events from your data warehouse).

Specification
  • 1.1OpenAPI 3.1 spec — generate clients in any language
  • 1.27 OAuth scopes — least-privilege by default
  • 1.3Per-token rate limits + usage analytics
  • 1.4Every API call audit-logged with request body + actor
API reference →
Phase 02

Webhooks · Signed events, exactly-once

Subscribe to events: new contact, agent action taken, deal stage changed, conversation started, anything in the platform. Payloads are signed (HMAC-SHA256), idempotent, and delivered with exponential backoff.

Most common wiring: webhook → your data warehouse → your BI layer. Or: webhook → Slack → your team gets pinged the moment the agent qualifies a high-intent lead.

Specification
  • 2.1HMAC-SHA256 signed payloads — verify in your handler
  • 2.2Idempotency keys + exponential backoff retry
  • 2.340+ event types covering every domain object
  • 2.4Per-event delivery log + replay endpoint
Webhooks reference →
Phase 03

SDKs · TypeScript, Python, Go

Official SDKs for the three languages we get asked about most. Auto-generated from the same OpenAPI spec, fully typed, with retry + pagination handled.

Community SDKs for Ruby + PHP exist (we link to them in the docs). If you'd prefer to use a raw HTTP client, that's fine — the REST API is the contract.

Specification
  • 3.1TypeScript SDK — npm, full types, ESM + CJS
  • 3.2Python SDK — PyPI, async + sync clients
  • 3.3Go SDK — pkg.go.dev, idiomatic structs + errors
  • 3.4Generated from OpenAPI — always in sync with API
SDK quickstarts →
Phase 04

Custom Skills · Author your own tools

Skills are the tools an agent can call mid-conversation — "look up this account in NetSuite," "check inventory in our warehouse system," "run our pricing model." You can author custom skills as tool manifests + an HTTPS endpoint we hit when the agent decides to call them.

Once registered, the skill is available to whichever agents you grant access. The agent decides when to call it based on context — you don't write the routing logic.

Specification
  • 4.1Tool manifest format — JSON schema + endpoint URL
  • 4.2Per-skill auth (mTLS, OAuth, signed JWT, API key)
  • 4.3Skill-level autonomy controls (auto / approval / off)
  • 4.4Test harness — preview agent decisions before going live
Skills reference →
FAQs

Frequently asked

Build on the platform we built ScendCore on.

Every primitive that runs our agents is open to your team. API, webhooks, SDKs, custom skills.

Built by a team that's delivered for
© 2026 ScendCore, operated by Sprintmore LLC · Frisco, Texas, USA
Developer Platform — Extend ScendCore with your code | ScendCore