Platform
Platform Overview Maps Memory Shadow Mode Widget Governance Actions
Developers
Overview Catalog Guardrails Dry Run Idempotency Credentials Audit Trail API Reference
Company
About Blog Contact Book a Demo
Runtime — Catalog

2,800+ actions.
One endpoint.

The Cerebral catalog is everything your agent can do. Pull it once — get the full action list as raw JSON Schema, model-agnostic and framework-agnostic by default. Add ?format=openai to get OpenAI function-calling blocks ready to drop straight into a tools[] array. Every action includes what it can call, what inputs it needs, and what risk level it carries.

2,800+
Actions in the catalog
Model-agnostic
JSON Schema by default
3-tier
Risk level on every verb
API Docs → Get API Access
The endpoint

One call.
Your agent knows everything it can do.

GET the catalog. Get raw JSON Schema for every action — model-agnostic, works with any framework. Add ?format=openai and each verb also gets a tool_definition block in OpenAI function-calling format, ready to drop into a tools array. Your agent now has access to 2,800+ actions — with inputs, outputs, and risk levels described for every one.

GET /v1/runtime/actions/catalog
# Pull the full catalog
GET /v1/runtime/actions/catalog
X-API-Key: sk_live_...
# Response — model-agnostic JSON Schema (default)
{
  "verbs": [{
    "verb": "shopify:get_order",
    "provider": "shopify",
    "risk_level": "low",
    "connected": true,
    "schema": {
      "type": "object",
      "properties": { ... }
    }
  }],
  "total": 17578
}
# Add ?format=openai for tool_definition blocks
  "tool_definition": {
    "type": "function",
    "function": { "name": "shopify__get_order", ... }
  }
Model-agnostic by default
The default response is raw JSON Schema — no LLM-specific envelope. Works with any model or framework. Add ?format=openai to append a tool_definition block per verb in OpenAI function-calling format.
Risk level on every verb
Every action in the catalog includes its risk_level — LOW, MEDIUM, or HIGH. Your agent can use this to decide whether to proceed or pre-flight check with /governance/evaluate.
Filter by provider or risk
Pass ?provider=shopify or ?risk=LOW to scope the catalog to what your agent needs. Reduces payload, speeds up tool list loading.
Tenant-scoped
The catalog returns only the verbs enabled for your tenant. Your agent only sees what it's allowed to call — governance starts at catalog time, not just execution time.
Risk classification

Every action has
a risk level. Always.

Risk classification is not something you configure. Every verb in the catalog ships with a pre-assigned risk level based on its write surface, reversibility, and potential impact. Your agent knows before it calls.

LOW
Executes automatically
Read operations, lookups, data fetches. No side effects. Always logged, never blocked.
shopify:get_order
stripe:retrieve_customer
hubspot:get_contact
MEDIUM
Executes with notification
State-changing but reversible. Executes immediately. Configurable for approval if your policy requires it.
shopify:update_order
klaviyo:tag_customer
reamaze:send_reply
HIGH
Blocked until approved
Irreversible or high-value actions. Blocked at the registry level until you configure an explicit approval gate.
stripe:create_refund
shopify:cancel_order
stripe:delete_customer
Provider coverage

If it has an API,
it's in the catalog.

2,800+ providers across e-commerce, CRM, payments, support, marketing, logistics, and more. Your agent doesn't need to know how to authenticate with any of them — credentials are resolved automatically at execution time.

ShopifyStripeHubSpot SalesforceKlaviyoRe:amaze GmailSlackJira ZendeskTwilioNotion AirtableNetSuitePayPal IntercomPipedriveMonday.com AsanaLinearMailchimp ActiveCampaignQuickBooksXero + 2,776 more

Your agent.
2,800+ actions.

One GET request. Full action catalog. Model-agnostic JSON Schema by default. Add ?format=openai for OpenAI tool definitions.

View API Docs ← Back to Developers