Daemon-AI Platform Skill

Daemon-AI (daemon-ai-backend) is an AI agent backend platform built on Express + TypeScript. It exposes an OpenAI-compatible API layered on top of OpenRouter, with blockchain tools, finance data, and USDC-based billing.

Base URL: https://daemon-ai-production.up.railway.app

Quick orientation

Who you are	What you need	Where to look
Developer integrating auth	Register, login, email verify	`references/public-endpoints.md`
Developer building an app	Chat, models, agent info, finance	`references/agent-endpoints.md`
Building an agentic loop	Tool use, streaming, MCP tools	`references/agentic-patterns.md`
AI agent calling the API	Auth flow + chat completions + tools	All three reference files

Authentication overview

There are two key auth mechanisms for public-facing work:

1. Agent API Key (`daemon_*`)

Used for all /v1/* endpoints (except auth registration/login which are public).

Authorization: Bearer daemon_<key>

Keys are prefixed with daemon_ and returned once at creation/first login. Store them securely — they cannot be retrieved again.

2. User Session Token

Returned on login. Used for user-management routes (/v1/agents).

Authorization: Bearer <session_token>

Session tokens expire after 30 days.

Endpoint groups

Group	Auth required	Reference
Health, Auth, Billing webhook	None	`references/public-endpoints.md`
Chat, Models, Agent info/settings	Agent API Key	`references/agent-endpoints.md`
Discover, Finance, Chat history	Agent API Key	`references/agent-endpoints.md`
MCP server management	Agent API Key	`references/agent-endpoints.md`
Agentic loop, streaming, tool use	—	`references/agentic-patterns.md`

Subscription tiers

Tier	Condition	Model access	Token limit
Free	Balance < $10 USDC	Whitelisted free models only	1M tokens/month
Paid (Pro)	Balance ≥ $10 USDC	All 300+ OpenRouter models	Custom (default 1M)

Free models include: arcee-ai/trinity-large-preview:free, stepfun/step-3.5-flash:free

Rate limits

Free tier: 10 requests / minute per API key
Paid tier: 60 requests / minute per API key

Common response patterns

Success:

{ "success": true, "data": { ... } }

Error:

{ "success": false, "error": "Human-readable message", "code": "MACHINE_CODE" }

Pagination:

{
  "data": [...],
  "pagination": { "page": 1, "limit": 20, "total": 100, "pages": 5 }
}

Reference files

Read these when you need full endpoint details:

references/public-endpoints.md — No-auth endpoints: health, register, login, wallet nonce, email verify, billing webhook
references/agent-endpoints.md — Agent API key endpoints: chat, models, agent info, finance, discover, chat history, MCP servers, API key management
references/agentic-patterns.md — How the agentic loop works, MCP tools available, streaming event format, tool execution, graph data

daemon-intelligence