Custom tools in TeamAI let your AI models and agents call external APIs and services. Build a tool once in your workspace, then turn it on in chat (Advanced Mode) or attach it to an agent.
Tools have replaced plugins in the TeamAI interface. All functionality previously available through custom plugins is now accessed through the Tools section.
Where to create a tool
1. Open Tools in the left navigation (wrench icon).
2. Click + Create a Tool.
3. Choose API Tool when you need GET or POST REST integrations.
For a full step-by-step walkthrough (variables, test, launch), see How to Use Tools (formerly called "plugins").
Note: Tools must be set up in your workspace before you can toggle them on in a chat conversation. The chat right panel only turns existing tools on or off for that conversation.
Tool types (overview)
Tool type | When to use |
API Tool | Connect to any REST API (GET or POST) |
MCP Server | Connect to Model Context Protocol servers |
Form Collection | Gather structured user input |
MCP Studio | Multi-step tool workflows |
This article focuses on API Tool configuration. For GET-specific and POST-specific setup, see the related articles at the bottom of this page.
Key fields in tool creation
When creating a custom tool, configure these fields on the Details tab:
Name
The name is how AI models identify and invoke the tool. Choose a clear, descriptive name that reflects what the tool does.
Description
The description gives models context about when and how to use the tool. Write a concise summary of the tool's purpose and expected inputs.
Icon (avatar)
The icon appears throughout TeamAI when the tool is listed or toggled. Pick an image that matches the tool's purpose, or use Generate with AI on the Details tab.
Endpoint and methods
On the API Setup tab, configure how the tool calls your API.
Endpoint
Enter the full API URL in API Connection, including any path segments. You can embed variables in the URL using double curly braces (see Variables below).
Methods
TeamAI API tools support GET and POST:
- GET: Retrieve data from an API (queries, lookups, calculations).
- POST: Send data in a request body (create, update, or submit operations).
For walkthroughs by method, see Creating a GET Route Custom Tool in TeamAI and Creating a POST Route Custom Tool in TeamAI.
Variables
Variables let the model supply dynamic values when it runs the tool.
What variables are
Variables are placeholders you define in tool configuration. The model fills them in based on the user's request. Reference a variable anywhere in the endpoint, headers, query parameters, or payload using `{{variableName}}`.
Creating variables
1. On the API Setup tab, open the Variables section.
2. Click Add Variable.
3. Enter a Variable Name and Description (the model uses both).
4. Select the type (text, number, boolean, or list types as needed).
5. Mark Required if the tool cannot run without that value.
6. Click Add Variable (or Update Variable) to save.
Using variables
Reference variables with double curly braces: `{{variableName}}`. Names must match exactly between the Variables list and wherever you use them (endpoint, headers, query parameters, or payload).
For naming patterns that help models pick the right tool, see Naming Conventions for Custom Tools in TeamAI.
Headers, query parameters, and payloads
Headers
Headers are key-value pairs sent with the request. Click Add Header in the Headers section, then enter the key and value. Use `{{variableName}}` in values when a value should be dynamic.
Common examples: `Authorization`, `Content-Type`, `Accept`.
Query parameters
Query parameters append to the URL as `?key=value`. Click Add Parameter in the Query Parameters section. Use variables in values the same way as in the endpoint.
Payloads (POST only)
For POST requests, define the request body in the payload section. Click Add Property (or the equivalent field control in your POST payload UI) and map keys to static values or `{{variableName}}` placeholders.
Hidden option for sensitive information
For headers, query parameters, and payload values that must stay secret (API keys, bearer tokens, passwords), enable Hidden on that row. Hidden values are not shown the next time you edit the tool.
Do not paste live secrets into shared or organization-visible tools without using Hidden or workspace secret management.
System variables
TeamAI provides built-in system variables (for example, user or workspace context) that you can reference in tool configuration. See Enhancing Custom Tools with System Variables.
Test before you use the tool
After configuration, open the Test & Launch tab, enter sample variable values, and click Run Test. Confirm the response before saving and using the tool in chat or on an agent.
Summary
Custom tools extend TeamAI by connecting models to external APIs. Configure name, description, and icon on Details; endpoint, method, variables, headers, and parameters on API Setup; then test on Test & Launch. Use meaningful names and descriptions, `{{variables}}` for dynamic values, and Hidden for secrets.