Skill: Flow Engine Avatar Control

This skill provides high-level control over a digital human avatar rendered by the Flow Engine (WebGPU). It allows agents to drive the avatar's speech, thought processes, and physical animations through a unified behavior protocol via MCP.

promptBeginner5 min to valuemarkdown

0 views

Feb 12, 2026

Skill: Flow Engine Avatar Control

Overview

Capabilities (Skills)

1. say

Description: Makes the avatar speak by displaying a speech bubble and playing the 'talking' animation.
Input Parameters:
- ```
text
```
  :
```
string
```
  , The text content to display in the bubble,
```
required
```
  .
- ```
duration
```
  :
```
number
```
  , How long the avatar stays in the talking state in milliseconds,
```
optional
```
  , default: 3000.
Output:
- ```
content
```
  :
```
array
```
  , Standard MCP content array confirming the action.

Example Usage:

{ "text": "Hello, I am Flow Engine!", "duration": 5000 }

2. think

Description: Places the avatar in a 'thinking' state, playing a thinking animation and showing a thought bubble.
Input Parameters:
- ```
text
```
  :
```
string
```
  , The content of the thought bubble,
```
optional
```
  , default: "...".
- ```
duration
```
  :
```
number
```
  , How long the avatar stays in the thinking state in milliseconds,
```
optional
```
  , default: 3000.
Output:
- ```
content
```
  :
```
array
```
  , Standard MCP content array confirming the action.

Example Usage:

{ "text": "Analyzing data...", "duration": 4000 }

3. play_action

Description: Triggers a specific pre-defined animation state on the avatar (e.g., waving, bowing).
Input Parameters:
- ```
action
```
  :
```
string
```
  , The name of the animation clip to play (e.g., 'wave', 'bow', 'dance'),
```
required
```
  .
Output:
- ```
content
```
  :
```
array
```
  , Standard MCP content array confirming the action.

Example Usage:

{ "action": "wave" }

Limitations

Requires a running WebGPU-capable browser environment for visual output.
Animation names depend on the specific
```
.glb
```
model and its configuration JSON.
Does not support real-time audio lip-sync in the current release.

Dependencies

@guoquan/flow-engine: current release, Core rendering and behavior engine.
THREE.js: r170+, Underlying 3D math and scene graph.
Zod: ^3.x, Schema validation and type safety.

Maintainers

Flow Engine Team - guoquan.net

View Original Source

Related Skills

Writing

PromptBeginner5 minmarkdown

English Translator and Improver

I want you to act as an English translator, spelling corrector and improver. I will speak to you in any language and you will detect the language, translate it and answer in the corrected and improved...

Jan 15, 2026

Writing

PromptBeginner5 minmarkdown

Bakery Merge Bounty Game Overview

Act as a Game Description Writer. You are responsible for crafting an engaging and informative overview of the mobile game '${gameName:Bake Merge Bounty}'. Your task is to highlight the core gameplay...

Jan 22, 2026

Writing

PromptBeginner5 minmarkdown

VEO3 Video Generation Assistant System Prompt

You are a specialized assistant for Google VEO3 video generation, designed to help users create, enhance, and optimize text prompts for high-quality video output. Your expertise encompasses the complete VEO3 prompting framework, technical specifications, and creative best practices.

Jan 15, 2026

Skill: Flow Engine Avatar Control

Description: Makes the avatar speak by displaying a speech bubble and playing the 'talking' animation.
Input Parameters:
- ```
text
```
  :
```
string
```
  , The text content to display in the bubble,
```
required
```
  .
- ```
duration
```
  :
```
number
```
  , How long the avatar stays in the talking state in milliseconds,
```
optional
```
  , default: 3000.
Output:
- ```
content
```
  :
```
array
```
  , Standard MCP content array confirming the action.

Example Usage:

{ "text": "Hello, I am Flow Engine!", "duration": 5000 }

2. think

Description: Places the avatar in a 'thinking' state, playing a thinking animation and showing a thought bubble.
Input Parameters:
- ```
text
```
  :
```
string
```
  , The content of the thought bubble,
```
optional
```
  , default: "...".
- ```
duration
```
  :
```
number
```
  , How long the avatar stays in the thinking state in milliseconds,
```
optional
```
  , default: 3000.
Output:
- ```
content
```
  :
```
array
```
  , Standard MCP content array confirming the action.

Example Usage:

{ "text": "Analyzing data...", "duration": 4000 }

3. play_action

Description: Triggers a specific pre-defined animation state on the avatar (e.g., waving, bowing).
Input Parameters:
- ```
action
```
  :
```
string
```
  , The name of the animation clip to play (e.g., 'wave', 'bow', 'dance'),
```
required
```
  .
Output:
- ```
content
```
  :
```
array
```
  , Standard MCP content array confirming the action.

Example Usage:

{ "action": "wave" }

Limitations

Requires a running WebGPU-capable browser environment for visual output.
Animation names depend on the specific
```
.glb
```
model and its configuration JSON.
Does not support real-time audio lip-sync in the current release.

Dependencies

@guoquan/flow-engine: current release, Core rendering and behavior engine.
THREE.js: r170+, Underlying 3D math and scene graph.
Zod: ^3.x, Schema validation and type safety.

Maintainers

Flow Engine Team - guoquan.net