User interface features
The developer console (/user/) lets you create and manage AI agents. Core features include Marketplace, History, My agents, Knowledge base, Memory and database, and Tools and skills, along with topic entry points such as FunClaw and FunHermes.
Marketplace
A discovery page for exploring high-quality agents, models, tools, and knowledge bases. The marketplace contains the following four tabs:
Tab | Description |
Agent | Displays available agent templates and examples. Use these templates to quickly create agent instances. |
Model | Browse the AI models available on the platform. |
Tool | Discover available tools and skills. |
Knowledge Base | Browse available knowledge base resources. |
History
View your session history, including session IDs, statistics, and timestamps. Built on OpenTelemetry, it provides observability data for monitoring agent health.
My agents
Manage your AgentRuntime instances. You can create two types of agents: conversational and orchestration-based.
Open Platform Admin Console developer console is available in China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China (Chengdu), and China (Hong Kong).
Supported operations:
Create a new AgentRuntime.
Copy an existing AgentRuntime.
Delete an AgentRuntime.
View an agent's access URL and runtime endpoint.
Agent instances are displayed in a card grid. You can search by name or description and view token usage for the current month. If you have no instances, the system guides you to create one.
Create an agent
On the My agents page, click the create button to create an agent.
Select an agent type
Select the type of agent you want to create:
Type | Description |
Conversational | Uses natural language conversation for multi-turn interactions. |
Orchestration-based | Uses workflow orchestration to automate complex business processes. |
One-shot AI generation | Quickly generate an agent from a single sentence (coming soon). |
Basic information
After selecting a type, configure the following:
Parameter | Description |
Agent name | Required. The system generates a default name automatically. |
Agent function description | A brief description of the agent's purpose and capabilities. |
After you complete the basic information, proceed to select a model, tools, skills, and more.
File management
Manage files in the agent's workspace for file transfers between the agent and the user. When a conversation involves files, you are directed to this page to download them.
Function | Description |
Browse | Browse the directory structure and file content of the workspace. |
Upload | Upload files to the workspace. |
Download | Download files from the workspace. |
Edit | Edit files in the workspace online. |
New | Create new files in the workspace. |
Knowledge base
Create and manage knowledge bases to enable intelligent Q&A on your private data by using Retrieval-Augmented Generation (RAG).
Function | Description |
Create knowledge base | Create a knowledge base by uploading files. |
External knowledge base connection | Connect to external knowledge base services like RAGFlow. |
Document management | View, delete, and refresh documents within a knowledge base. |
Knowledge base list | Displays the name, tags, latest version, and number of associated agents for each knowledge base. |
Prompt management
Create, edit, and manage prompts with version management and debugging capabilities.
Function | Description |
Create and edit | Create a new prompt or edit an existing one. |
Version management | Track the version history of a prompt. |
Debugging and comparison testing | Use debugging to test prompt effects directly. Use comparison testing to run multiple versions with different models or parameters simultaneously and find the optimal configuration. |
Prompt variables | Define variable names, descriptions, and values for reuse and tuning across scenarios. |
Memory and database
Manage an agent's long-term memory and database storage:
Create memory store: Configure an independent LLM and embedding model.
Manage long-term and short-term memory. Long-term memory persists across sessions. Short-term memory lasts only for the current session and is cleared automatically when the session ends.
View and delete memory entries.
Tools and skills
Register and manage external capabilities that an agent can call. Five filter tabs are available: All, Function Call, MCP, Skill, and Composite Tool.
Tool type | Description |
Function Call | Register custom function call tools to extend agent capabilities. |
MCP | Register an MCP (Model Context Protocol) tool. |
Skill | Register a skill. You can create one by using a Markdown definition or by uploading a skill package. |
Composite Tool | Combine multiple tools into a single composite tool to simplify the call process. |
Workspace
A workspace is an independent unit for team collaboration. Workspaces are managed from the drop-down switcher in the top navigation bar, not from a sidebar page.
Function | Description |
Create workspace | Create an independent workspace. |
Member management | View the member list, search for members, and manage roles (administrator or user). |
Apply to join | Apply to join another workspace. |
Approval | Workspace administrators approve join requests. |
Topics
The Topics section in the sidebar provides the following entry points:
FunClaw: Entry point for OpenClaw-related features. For more information, see Topic-FunClaw.
Help
The bottom of the sidebar links to the following help pages:
Quick start: Get started with the platform.
Use cases: Typical use cases and examples.
Best practices: Recommended methods and practices.
API documentation: API reference.
FAQ: Frequently Asked Questions.
Profile
Access this section from the user avatar menu in the top navigation bar. It includes the following features:
Basic information: View and edit your personal profile.
Account security: Change your login password.