# GreenNode AgentBase

## Start with a real example

You want to build a customer support AI Agent: it takes questions via chat, looks up orders in a database, sends notifications through Slack, and remembers what was discussed in last week's conversation.

Sounds straightforward — but in practice you'll need to handle:

* **Where does the agent run?** Containers, servers, autoscaling, CI/CD deployments...
* **Where do credentials live?** Database passwords, Slack tokens, API keys — you can't hardcode them.
* **Can the agent call any tool it wants?** You need guardrails to prevent it from accidentally calling a data-deletion API.
* **What did the LLM cost this month?** No dashboard, no way to know when you're about to exceed budget.
* **When something breaks in production?** No centralized logs, no visibility into which requests failed.

**AgentBase handles all of this** — so you can focus entirely on your agent's logic.

***

## What is AgentBase?

**GreenNode AgentBase** is a purpose-built infrastructure platform for AI Agents — providing the full operational, security, and governance layer needed to take an agent from code to production.

AgentBase consists of the following modules:

| Module                 | Function                                                                                                         |
| ---------------------- | ---------------------------------------------------------------------------------------------------------------- |
| **Agent Runtime**      | Deploy and operate agents — container lifecycle management, versioning, rollback, scaling                        |
| **Marketplace**        | Deploy pre-built agents (OpenClaw and templates) with 1 click, no code required                                  |
| **Access Control**     | Manage Agent Identity and store credentials (API Key, OAuth2) — automatically injected into the agent at runtime |
| **MCP Governance**     | Control all MCP tool calls from agents — authentication and authorization via MCP Gateway + Policy Group         |
| **Protect & Govern**   | Rate Limiting by model or API Key — prevent agents from consuming excessive resources                            |
| **Memory**             | Give agents cross-session memory — Short-Term (conversation history) and Long-Term (semantic search)             |
| **Container Registry** | Private image registry automatically created per org — stores container images for Custom Agents                 |
| **Team & Permissions** | Manage members with 4 roles (Root / Admin / Member / Viewer) and granular permission control                     |
| **Usage & Budget**     | Dashboard tracking requests, tokens, and cost by agent/model/provider; set budget limits and automated alerts    |

***

## Two ways to get started

**No code — use immediately:** Go to the [Marketplace](/vng-cloud-document/ai-stack/agent-base/marketplace.md), select **OpenClaw**, enter your API key and chat channel — the agent is running in minutes.

**Build your own agent:** Package your agent as a Docker image, push it to [Container Registry](/vng-cloud-document/ai-stack/agent-base/container-registry.md), and deploy via [Agent Runtime](/vng-cloud-document/ai-stack/agent-base/agent-runtime.md). Add credentials in [Access Control](/vng-cloud-document/ai-stack/agent-base.md), and attach an [MCP Gateway](/vng-cloud-document/ai-stack/agent-base/mcp-governance/mcp-gateway.md) if the agent needs to call external tools.

***

## Who is it for?

| Audience                      | What AgentBase delivers                                                   |
| ----------------------------- | ------------------------------------------------------------------------- |
| **AI Engineers / Developers** | Focus on agent logic — infra, credentials, and observability are built in |
| **Startups / Product Teams**  | Ship AI products faster.                                                  |
| **Enterprises**               | Cost control, team-based access, enterprise-grade credential security     |


---

# 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.vngcloud.vn/vng-cloud-document/ai-stack/agent-base.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.