# Agent System

Noelclaw's agent marketplace hosts 40+ specialized AI agents. Agents can be chatted with directly, automated via schedules, or accessed externally via MCP.

***

## Agent Types

| Badge    | Meaning                                                            |
| -------- | ------------------------------------------------------------------ |
| `mcp`    | Accessible via MCP protocol (external clients like Hermes, Claude) |
| `x402`   | Requires x402 micropayment to use                                  |
| `acp`    | Anthropic Cloud Provider — uses Claude models                      |
| `plugin` | Data feed plugin (CoinGecko, Bankr, etc.)                          |

***

## Featured Agents

### Noel

> Crypto intelligence agent with DeFi trading expertise

* **Model:** gpt-5-nano (via Bankr gateway)
* **Specialty:** Market analysis, trade ideas, DeFi research, on-chain signals
* **Pricing:** 0.001 USDC per action
* **MCP tool:** `ask_noel`
* **Research:** Runs autonomous 8h research shifts (see [Research System](/core-features/research-system.md))

### Gloria AI

> AI/Web3 news and intelligence agent

* **Model:** claude-sonnet-4-6
* **Specialty:** AI agent news, Web3 ecosystem trends, project analysis, headlines
* **MCP tool:** Coming soon
* **Status:** Active in chat, MCP integration planned

### CoinGecko Agent

> Live token market data

* **Model:** claude-sonnet-4-6
* **Specialty:** Token prices, 24h changes, market cap, volume
* **MCP tool:** `get_token_data`
* **Source:** CoinGecko API

### Bankr

> On-chain signals and DeFi intelligence

* **Specialty:** Whale movements, smart money tracking, on-chain analytics
* **Integration:** Used as LLM gateway + data source for Noel's research

### Loky

> Technical analysis agent

* **Specialty:** Chart patterns, support/resistance, TA signals

### Whale Intel

> Large wallet monitoring

* **Specialty:** Whale wallet tracking, accumulation alerts

### AIXBT

> AI-native alpha agent

* **Status:** Coming soon

***

## Custom Agents

Users can create their own agents in **Create Agent** page.

**Configuration options:**

* Name, description, avatar
* Category (DeFi, News, Data, Code, etc.)
* System prompt (defines behavior)
* Model provider: Claude, ChatGPT, Gemini, Kimi, Qwen, Bankr, Custom
* Pricing: Free or token-based (USDC per action)
* Voice ID (for voice mode)
* Publish to marketplace toggle

**Database:**

```
customAgents table:
  userId, name, description, avatar
  category, systemPrompt
  voiceId, pricingType, modelId
  isPublished, rating, reviewCount
  runs, activeUsers
```

***

## Agent Chat Interface

When a user opens a chat with any agent:

1. System prompt is loaded from agent config
2. User messages go to `api.chat.chat` (Convex action)
3. Convex calls Bankr LLM gateway with the model
4. Response streams back to the UI
5. Session saved to memory (searchable later)

**Research-enabled chat:** If the user has "Noel Research" skill enabled, each chat message triggers the research orchestrator first — injecting live market context before Claude responds.

***

## Agent Runs Log

Every skill execution is logged in `agentRuns`:

```
agentRuns table:
  userId      — who triggered it
  skillName   — e.g., "noel-research"
  trigger     — "schedule" | "onDemand" | "event"
  status      — "running" | "done" | "error"
  tokensUsed  — total tokens consumed
  cost        — estimated USDC cost
  result      — JSON result object
  error       — error message if failed
  startedAt   — Unix timestamp
  finishedAt  — Unix timestamp
```

These are visible in the **Brain** page and **Memory** tab.

***

## Skill Configuration

Users install and configure skills in **Automations** page:

```
userSkillConfig table:
  userId    — owner
  skillName — e.g., "noel-research"
  config    — JSON (interval, preferences)
  enabled   — boolean
  installedAt — timestamp
```

Available skills:

* `noel-research` — 8h autonomous research
* `gloria-default` — AI news updates
* `coingecko-default` — Market data fetches


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.noelclaw.fun/core-features/agents.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.