> ## Documentation Index
> Fetch the complete documentation index at: https://docs.chatropic.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Actions overview

> How your agent's tools are created from skills and managed

Actions are the tools your agent invokes when a conversation matches their description: an API lookup, a form, a widget, or an escalation. You do not build actions by hand. **All actions are created from [Skills](/user-guides/agent-management/skills)**: when you save a skill that contains a webhook step, Chatropic creates the linked HTTP action and lists it under **Actions > Available actions**.

<Frame caption="The Available actions list under Actions.">
  <img src="https://mintcdn.com/chatropic/VGIVt7hCVdFIOyKi/images/Screenshot-2026-07-03-at-2.12.06-PM.png?fit=max&auto=format&n=VGIVt7hCVdFIOyKi&q=85&s=5d0725c74af74cbf52006f630a8bdad7" alt="Available actions list" lightAlt="Available actions list" darkAlt="Available actions list" className="dark:hidden" width="3420" height="1804" data-path="images/Screenshot-2026-07-03-at-2.12.06-PM.png" />

  <img src="https://mintcdn.com/chatropic/VGIVt7hCVdFIOyKi/images/Screenshot-2026-07-03-at-2.12.06-PM-1.png?fit=max&auto=format&n=VGIVt7hCVdFIOyKi&q=85&s=539395552051ed6c9704f25161aa6e11" alt="Available actions list" lightAlt="Available actions list" darkAlt="Available actions list" className="hidden dark:block" width="3420" height="1804" data-path="images/Screenshot-2026-07-03-at-2.12.06-PM-1.png" />
</Frame>

## What is an action?

An action defines:

| Field            | Role                                                                          |
| ---------------- | ----------------------------------------------------------------------------- |
| **Display name** | Human-readable label in the dashboard                                         |
| **Action ID**    | Snake-case identifier used in chaining (auto-generated from the display name) |
| **Description**  | When to call, read by the planner each turn                                   |
| **Parameters**   | Fields the agent collects from the user                                       |
| **Config**       | HTTP endpoint, widget mapping, or response mapping                            |
| **Enabled**      | Toggle without deleting                                                       |

```mermaid theme={null}
flowchart LR
    U[User message] --> P[Planner]
    P -->|matches description| A[Action]
    A --> API[Your API or widget]
    API --> R[Reply + optional UI]
```

## Where actions come from

<Steps>
  <Step title="Create a skill">
    Describe the feature under **Agent settings > Skills > New skill**, including the API call if one is needed. See the [Skills guide](/user-guides/agent-management/skills).
  </Step>

  <Step title="Save the skill">
    Each webhook step in the skill creates (or updates) a linked HTTP action automatically.
  </Step>

  <Step title="Review in Available actions">
    Open **Actions > Available actions** to see the linked action, test it with sample parameters, and enable or disable it.
  </Step>
</Steps>

<Note>
  Edit an action's URL, headers, and body from its skill's webhook step rather than from the action page, so the skill and its linked action stay in sync.
</Note>

## How the agent fills parameters at runtime

1. The planner selects an action when the user's message matches the action **Description**.
2. It fills the parameters from the message and prior turns, using the property names from the parameter schema.
3. Chatropic substitutes `{{parameter}}` placeholders in the URL and body before calling your API.
4. If a **required** parameter is missing, the agent asks a follow-up question instead of calling the API with an empty value.

This is why the description matters so much: write it like an instruction to a colleague, for example *"Search the catalog by keyword when the user asks about items, prices, or availability."*

## Go deeper

<CardGroup cols={2}>
  <Card title="Skills" icon="wand-magic-sparkles" href="/user-guides/agent-management/skills">
    Create the playbooks that generate actions.
  </Card>

  <Card title="Manage tokens" icon="key" href="/user-guides/actions/manage-tokens">
    Static and customer-login credentials for API auth.
  </Card>

  <Card title="Action chaining" icon="link" href="/user-guides/actions/action-chaining">
    Multi-step workflows with on\_success.
  </Card>

  <Card title="Response mapping" icon="arrow-right-arrow-left" href="/user-guides/actions/response-mapping">
    Shape API JSON for chat widgets.
  </Card>

  <Card title="Escalate to human" icon="headset" href="/user-guides/actions/escalate-to-human">
    Hand off to a live agent queue.
  </Card>

  <Card title="Suggested prompts" icon="lightbulb" href="/user-guides/actions/suggested-prompts">
    Quick-reply chips and starter prompts.
  </Card>
</CardGroup>

<Warning>
  Do not put secrets in action URLs or headers. Use [Manage tokens](/user-guides/actions/manage-tokens) and reference them with `@` in header fields.
</Warning>
