Documentation

Agents

Agents are named configurations (slug, model tier, system prompt) that you dispatch by id on every run.

Default catalog

Each workspace receives six starter agents on onboarding. Slugs are stable API identifiers.

Slug	Tier	Billing	Purpose
quick-ask	fast	free	Playground Q&A, smoke tests
wallet-audit	frontier	metered	On-chain wallet summary JSON
morning-brief	balanced	metered	Daily digest structure
launch-scout	frontier	metered	Launch scan cards
token-scout	balanced	metered	Single-token dossier
risk-scan	fast	metered	Pass/fail risk gates

Input contract

Runs accept JSON input. All agents support at minimum:

Standard message input

{
  "message": "Your user or integrator prompt here"
}

Wallet audit

Include a checksummed address so the orchestrator can attach verified native balance context before inference:

wallet-audit

{
  "message": "Audit my exposure",
  "wallet_address": "0x742d35Cc6634C0532925a3b844Bc454e4438f44e"
}

Grounding

Structured agents return JSON with data_status and missing_data when live feeds are not present — they do not invent holdings or market numbers.

Custom agents

Create agents in the console or via API. Custom agents are always metered (prepaid credits required).

POST /v1/agents

POST https://runlayer-y9kc.onrender.com/v1/agents
Authorization: Bearer <token-or-key>
{
  "name": "Support triage",
  "slug": "support-triage",
  "model_tier": "balanced",
  "system_prompt": "You classify tickets into billing, technical, or sales.",
  "description": "Internal L1 triage"
}

List agents

GET /v1/agents

GET https://runlayer-y9kc.onrender.com/v1/agents
Authorization: Bearer <token-or-key>

billingMode: free or metered
model: fast | balanced | frontier — maps to inference routing