Gateway Endpoints

Gateway routes are mounted under /api/gateway/* and enforce auth/billing checks before upstream proxying.

LLM Gateway

Base route: /api/gateway/llm

Method	Path
POST	/api/gateway/llm/v1/chat/completions
POST	/api/gateway/llm/v1/responses

Search Gateway

Base route: /api/gateway/search

Method	Path
GET	/api/gateway/search/v1/web/search
GET	/api/gateway/search/v1/news/search

Usage Webhook and Usage View

Base route: /api/webhooks/usage

Method	Path	Description
POST	/api/webhooks/usage	ingest usage events
GET	/api/webhooks/usage	read usage snapshot/records

Copy/Paste Examples

LLM Proxy Request

curl -X POST "$ROBERTO_API_URL/api/gateway/llm/v1/chat/completions" \
  -H "x-api-key: $ROBERTO_AGENT_GATEWAY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "moonshotai/kimi-k2.5",
    "messages": [
      {"role":"user","content":"Summarize today'\''s support incidents in 3 bullets."}
    ]
  }'

Response shape is provider pass-through (OpenAI-compatible object).

Search Proxy Request

curl "$ROBERTO_API_URL/api/gateway/search/v1/web/search?q=robertoagent%20updates" \
  -H "Authorization: Bearer $ROBERTO_AGENT_GATEWAY_TOKEN"

Response shape is upstream search-provider pass-through.

Usage Webhook Ingest

curl -X POST "$ROBERTO_API_URL/api/webhooks/usage?agentId=$AGENT_ID" \
  -H "Content-Type: application/json" \
  -H "x-idempotency-key: evt_123" \
  -d '{
    "messages": 2,
    "inputTokens": 540,
    "outputTokens": 180,
    "eventId": "evt_123"
  }'

Example 200 response:

{
  "success": true,
  "duplicate": false
}

Read Usage Snapshot

curl "$ROBERTO_API_URL/api/webhooks/usage?agentId=$AGENT_ID"

Example 200 response:

{
  "messages": 120,
  "inputTokens": 45200,
  "outputTokens": 19100,
  "totalTokens": 64300,
  "estimatedCost": "0.2345",
  "estimatedCostUsd": 0.2345,
  "eventCount": 85,
  "lastUpdated": "2026-02-21T12:00:00.000Z"
}