Store
Store is where you find useful building blocks for your Agent.
Browse the live catalogs: Models, APIs, Agents, and Skills.
You can start from four kinds of things:
| Kind | What it is | What you do with it |
|---|---|---|
| Models | AI models for text, image, audio, and video | Call them directly, or use them inside an Agent |
| APIs | Ready-to-use capabilities like search, scrape, data, media, and SaaS actions | Copy a request or add to Setup |
| Agents | Proven Agent examples and best-practice workflows | Clone them, test them, and deploy your copy as a Service or Schedule |
| Skills | Reusable task packages for Agents | Attach them when building an Agent |
How to choose
If you only need one capability, start with Models or APIs.
If you want a complete example, start with Agents.
If you are building your own Agent and want reusable implementation knowledge, use Skills.
Decision guide
| Your goal | Start with | Then |
|---|---|---|
| Add one capability to an AI tool | Model or API | Add it to Setup |
| Call a capability from code | Model or API | Open its detail page, then use the API Reference |
| Automate a multi-step job | Agent | Clone, test, and deploy it from Build Agent |
| Reuse a proven workflow | Skill | Attach it while building an Agent |
When two entries look similar, compare their input, output, pricing, latency, and authentication requirements before installing or cloning them.
Models
Models are AI engines for text generation and reasoning, image generation or editing, audio, and video.
You can call a Model directly, add it to Setup for a connected AI tool, or select it when building an Agent. The Model determines how an Agent thinks; Skills and MCP tools determine what it can do.
Most text Models use the OpenAI-compatible Chat Completions API. See Supported Models and the Model capability matrix.
APIs
APIs give your application or connected AI tools access to real-world data and actions, including web search, scraping, business data, media generation, SaaS tools, and provider-specific capabilities.
Use an API in one of two ways:
- call it directly from your application
- add it to Setup so a connected AI tool can use it
Agents reach external systems through MCP connections.
Models think and generate; APIs fetch data, take actions, or connect external systems. See the API Reference for request formats and endpoints.
Agents
Store Agents are working examples that combine a Model, Skills, and workflow instructions.
You can inspect a public Agent, clone it into your workspace, customize it, and deploy your copy as a Service or Schedule. Clone the public Agent first when you need to change its instructions, model, Skills, MCP tools, or output format.
Skills
Skills are reusable task packages for Agents. They can contain instructions, workflows, code, assets, and repeatable know-how.
Use a Skill when an Agent needs consistent expertise, such as company research, code review, structured reporting, page analysis, or a customer-support workflow. Skills are attached in Build Agent; they are not top-level Setup tools.
Common paths
Use it in your AI app
Add Models, APIs, or Services to Workspace Services, then follow the client-specific flow in Setup. Supported clients and prerequisites are shown in the Console and can change independently of this overview.
Call it from your app
Open a Model or API detail page, copy a working request, and call it with your SandBase key.
Build your own Agent
Clone an Agent example or start from scratch in Build Agent. Choose a Model, attach Skills and MCP connections, test it in a Session, then deploy it as a Service or Schedule.
Before you add an item
Check the detail page for:
- the exact input and output shape
- pricing and whether usage is per token or per call
- required API credentials
- whether the item is public, maintained by SandBase, or owned by your workspace
- examples you can run before adopting it
