feat: Add intelligent auto-router and enhanced integrations
- Add intelligent-router.sh hook for automatic agent routing - Add AUTO-TRIGGER-SUMMARY.md documentation - Add FINAL-INTEGRATION-SUMMARY.md documentation - Complete Prometheus integration (6 commands + 4 tools) - Complete Dexto integration (12 commands + 5 tools) - Enhanced Ralph with access to all agents - Fix /clawd command (removed disable-model-invocation) - Update hooks.json to v5 with intelligent routing - 291 total skills now available - All 21 commands with automatic routing 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
admin
73
dexto/docs/examples/amazon-shopping.md
Normal file
73
dexto/docs/examples/amazon-shopping.md
Normal file
@@ -0,0 +1,73 @@
|
||||
---
|
||||
title: "Browser Agent: Amazon Shopping Assistant"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Browser Agent: Amazon Shopping Assistant
|
||||
|
||||
Automate web browsing tasks like shopping, research, and data collection with AI-powered browser control.
|
||||
|
||||
**Task:** `Can you go to amazon and add some snacks to my cart? I like trail mix, cheetos and maybe surprise me with something else?`
|
||||
|
||||
```bash
|
||||
# Default agent has browser tools
|
||||
dexto
|
||||
```
|
||||
|
||||
<a href="https://youtu.be/C-Z0aVbl4Ik">
|
||||
<ExpandableImage src="https://github.com/user-attachments/assets/3f5be5e2-7a55-4093-a071-8c52f1a83ba3" alt="Dexto: Amazon shopping agent demo" title="Browser Agent: Amazon Shopping Assistant" width={900} />
|
||||
</a>
|
||||
|
||||
## What it does
|
||||
|
||||
The default Dexto agent includes browser automation tools powered by Puppeteer:
|
||||
- Navigate websites
|
||||
- Fill out forms
|
||||
- Click buttons and links
|
||||
- Extract information
|
||||
- Make purchases (with approval)
|
||||
- Screenshot and analyze pages
|
||||
|
||||
## How it works
|
||||
|
||||
The agent uses browser tools to:
|
||||
1. Open websites in a real browser
|
||||
2. Understand page content
|
||||
3. Interact with elements (click, type, scroll)
|
||||
4. Complete multi-step tasks
|
||||
5. Return results or confirmations
|
||||
|
||||
## Use Cases
|
||||
|
||||
- **Shopping**: Find and add items to cart
|
||||
- **Research**: Collect data from multiple sources
|
||||
- **Form Filling**: Automate repetitive data entry
|
||||
- **Price Comparison**: Check prices across sites
|
||||
- **Booking**: Reserve hotels, flights, restaurants
|
||||
- **Content Extraction**: Scrape information from websites
|
||||
|
||||
## Try it
|
||||
|
||||
```bash
|
||||
# Use default agent (includes browser tools)
|
||||
dexto
|
||||
|
||||
# Example prompts
|
||||
"Find the cheapest wireless mouse on Amazon"
|
||||
"Book a table at an Italian restaurant nearby for 2 people at 7pm"
|
||||
"Compare prices for iPhone 15 on Best Buy and Amazon"
|
||||
```
|
||||
|
||||
## Safety Features
|
||||
|
||||
- **Approval prompts** before purchases or sensitive actions
|
||||
- **Session isolation** for security
|
||||
- **Headless or visible mode** options
|
||||
- **Screenshot capture** for verification
|
||||
|
||||
## Learn More
|
||||
|
||||
- [CLI Guide](/docs/guides/cli/overview)
|
||||
- [Agent Configuration](/docs/guides/configuring-dexto/overview)
|
||||
- [Tool Confirmation](/docs/guides/configuring-dexto/agent-yml#tool-confirmation)
|
||||
84
dexto/docs/examples/email-slack.md
Normal file
84
dexto/docs/examples/email-slack.md
Normal file
@@ -0,0 +1,84 @@
|
||||
---
|
||||
title: "Email to Slack: Automated Email Summaries"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Email to Slack: Automated Email Summaries
|
||||
|
||||
Automatically summarize emails and send highlights to Slack channels.
|
||||
|
||||
**Task:** `Summarize emails and send highlights to Slack`
|
||||
|
||||
```bash
|
||||
dexto --agent ./agents/examples/email_slack.yml
|
||||
```
|
||||
|
||||
<ExpandableImage src="/assets/email_slack_demo.gif" alt="Email to Slack Demo" title="Email to Slack: Automated Email Summaries" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
This example demonstrates multi-tool orchestration:
|
||||
1. Connect to email via Gmail MCP server
|
||||
2. Fetch recent unread emails
|
||||
3. Summarize content using LLM
|
||||
4. Send formatted summaries to Slack channel
|
||||
5. Mark emails as read
|
||||
|
||||
## Requirements
|
||||
|
||||
- Gmail access via Composio (SSE endpoint)
|
||||
- Slack MCP server (`@modelcontextprotocol/server-slack`)
|
||||
- Composio API setup for Gmail
|
||||
- Slack bot token
|
||||
- Agent configuration file
|
||||
|
||||
## Setup
|
||||
|
||||
1. **Configure Gmail access:**
|
||||
- Set up Composio for Gmail integration
|
||||
- Get your Composio endpoint URL
|
||||
- Configure authentication
|
||||
|
||||
2. **Get Slack token:**
|
||||
- Create a Slack app
|
||||
- Add bot token scopes: `chat:write`, `channels:read`
|
||||
- Set `SLACK_BOT_TOKEN` and `SLACK_TEAM_ID` environment variables
|
||||
|
||||
3. **Configure agent:**
|
||||
```yaml
|
||||
# agents/examples/email_slack.yml
|
||||
mcpServers:
|
||||
gmail:
|
||||
type: http
|
||||
url: "your-composio-url"
|
||||
|
||||
slack:
|
||||
type: stdio
|
||||
command: npx
|
||||
args: ['-y', '@modelcontextprotocol/server-slack']
|
||||
env:
|
||||
SLACK_BOT_TOKEN: $SLACK_BOT_TOKEN
|
||||
SLACK_TEAM_ID: $SLACK_TEAM_ID
|
||||
```
|
||||
|
||||
4. **Run the agent:**
|
||||
```bash
|
||||
dexto --agent ./agents/examples/email_slack.yml
|
||||
```
|
||||
|
||||
## Customization
|
||||
|
||||
Modify the agent to:
|
||||
- Filter emails by sender or subject
|
||||
- Custom summary formats
|
||||
- Schedule periodic checks
|
||||
- Route to different Slack channels based on content
|
||||
- Add reactions or threading
|
||||
|
||||
## Learn More
|
||||
|
||||
- [MCP Slack Server](https://github.com/modelcontextprotocol/servers/tree/main/src/slack)
|
||||
- [Composio Integration](https://composio.dev/)
|
||||
- [Agent Configuration](/docs/guides/configuring-dexto/overview)
|
||||
- [MCP Integration Guide](/docs/mcp/overview)
|
||||
45
dexto/docs/examples/face-detection.md
Normal file
45
dexto/docs/examples/face-detection.md
Normal file
@@ -0,0 +1,45 @@
|
||||
---
|
||||
title: "Computer Vision: Face Detection & Annotation"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Computer Vision: Face Detection & Annotation Using OpenCV
|
||||
|
||||
Detect faces in images and annotate them with bounding boxes using computer vision.
|
||||
|
||||
**Task:** `Detect all faces in this image and draw bounding boxes around them.`
|
||||
|
||||
```bash
|
||||
dexto --agent image-editor-agent
|
||||
```
|
||||
|
||||
<ExpandableImage src="https://github.com/user-attachments/assets/7e4b2043-c39a-47c7-a403-a9665ee762ce" alt="Face Detection Demo" title="Computer Vision: Face Detection & Annotation" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
The Image Editor Agent includes computer vision capabilities powered by OpenCV:
|
||||
- Detect faces in uploaded images
|
||||
- Draw bounding boxes with customizable colors
|
||||
- Apply filters and transformations
|
||||
- Save annotated results
|
||||
|
||||
## Requirements
|
||||
|
||||
- OpenAI GPT-5 Mini (or compatible model)
|
||||
- Image upload capability (Web UI or API)
|
||||
|
||||
## Try it
|
||||
|
||||
```bash
|
||||
# Install the agent
|
||||
dexto install image-editor-agent
|
||||
|
||||
# Run it
|
||||
dexto --agent image-editor-agent
|
||||
```
|
||||
|
||||
Upload an image in the Web UI, then ask:
|
||||
```
|
||||
"Detect all faces in this image and draw bounding boxes around them"
|
||||
```
|
||||
77
dexto/docs/examples/human-in-loop.md
Normal file
77
dexto/docs/examples/human-in-loop.md
Normal file
@@ -0,0 +1,77 @@
|
||||
---
|
||||
title: "Human In The Loop: Dynamic Form Generation"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Human In The Loop: Dynamic Form Generation
|
||||
|
||||
Agents can generate structured forms when they need additional data, making it easier to collect extra info and approvals from users.
|
||||
|
||||
<ExpandableImage src="/assets/user_form_demo.gif" alt="User Form Demo" title="Human In The Loop: Dynamic Form Generation" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
When an agent needs clarification or additional input, it can:
|
||||
- **Generate dynamic forms** with appropriate fields
|
||||
- **Validate user input** before proceeding
|
||||
- **Request approvals** for sensitive operations
|
||||
- **Collect structured data** in an intuitive way
|
||||
|
||||
## How it works
|
||||
|
||||
The agent automatically triggers form generation when it needs more information:
|
||||
|
||||
```bash
|
||||
# Example: Booking a flight
|
||||
> "Book me a flight to New York"
|
||||
|
||||
# Agent generates a form requesting:
|
||||
- Departure date
|
||||
- Return date
|
||||
- Preferred airline
|
||||
- Budget range
|
||||
- Seat preference
|
||||
```
|
||||
|
||||
You fill out the form, submit it, and the agent continues with the complete information.
|
||||
|
||||
## Use Cases
|
||||
|
||||
### 1. Tool Approvals
|
||||
Before executing sensitive operations (deleting files, making API calls), the agent requests confirmation with details about what will happen.
|
||||
|
||||
### 2. Missing Parameters
|
||||
When a task requires specific data the agent doesn't have, it generates a form to collect it efficiently.
|
||||
|
||||
### 3. Configuration
|
||||
Setting up complex configurations becomes easier with guided form inputs instead of free-form text.
|
||||
|
||||
### 4. Data Collection
|
||||
Collect structured information for reports, bookings, or any multi-field data entry.
|
||||
|
||||
## Configuration
|
||||
|
||||
Configure approval requirements in your `agent.yml`:
|
||||
|
||||
```yaml
|
||||
toolApproval:
|
||||
mode: selective
|
||||
requireApprovalFor:
|
||||
- deleteFile
|
||||
- executeCommand
|
||||
- makePurchase
|
||||
```
|
||||
|
||||
## Benefits
|
||||
|
||||
- **Better UX**: Structured forms are easier than back-and-forth messages
|
||||
- **Validation**: Ensure data is correct before processing
|
||||
- **Safety**: Explicit approvals for dangerous operations
|
||||
- **Efficiency**: Collect multiple fields at once
|
||||
|
||||
## Learn More
|
||||
|
||||
- [Agent Configuration](/docs/guides/configuring-dexto/overview)
|
||||
- [Tool Confirmation Settings](/docs/guides/configuring-dexto/agent-yml#tool-confirmation)
|
||||
- [MCP Elicitation](/docs/mcp/elicitation)
|
||||
81
dexto/docs/examples/image-generation.md
Normal file
81
dexto/docs/examples/image-generation.md
Normal file
@@ -0,0 +1,81 @@
|
||||
---
|
||||
title: "Hugging Face: Image Generation"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Hugging Face: Image Generation
|
||||
|
||||
Generate images using Hugging Face models with simple text prompts.
|
||||
|
||||
**Task:** `Generate a photo of a baby panda.`
|
||||
|
||||
```bash
|
||||
dexto --agent nano-banana-agent
|
||||
```
|
||||
|
||||
<ExpandableImage src="https://github.com/user-attachments/assets/570cbd3a-6990-43c5-b355-2b549a4ee6b3" alt="Hugging Face Image Generation Demo" title="Hugging Face: Image Generation" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
The Nano Banana Agent uses Google's Gemini 2.5 Flash Image model (formerly Nano Banana) for advanced image operations:
|
||||
- Generate images from text descriptions
|
||||
- Edit existing images
|
||||
- Apply style transformations
|
||||
- Create variations
|
||||
- Enhance image quality
|
||||
|
||||
## Requirements
|
||||
|
||||
- `GOOGLE_GENERATIVE_AI_API_KEY` environment variable
|
||||
- Google Gemini 2.5 Flash Image model (included in agent config)
|
||||
|
||||
## Try it
|
||||
|
||||
```bash
|
||||
# Install the agent
|
||||
dexto install nano-banana-agent
|
||||
|
||||
# Open the agent in web UI
|
||||
dexto --agent nano-banana-agent
|
||||
```
|
||||
|
||||
Try different Prompts to generate images:
|
||||
```
|
||||
"create a futuristic cityscape with flying cars"
|
||||
"generate a watercolor painting of a sunset over mountains"
|
||||
"create a cute robot mascot for a tech startup"
|
||||
```
|
||||
|
||||
## Features
|
||||
|
||||
- **High Quality**: Generated using state-of-the-art Google Gemini models
|
||||
- **Fast Generation**: Optimized for quick results
|
||||
- **Flexible Prompts**: Natural language descriptions
|
||||
- **Multiple Styles**: From photorealistic to artistic
|
||||
- **Batch Generation**: Create multiple variations
|
||||
|
||||
## Advanced Usage
|
||||
|
||||
### Style Control
|
||||
```
|
||||
"Generate a baby panda in watercolor style"
|
||||
"Create a photorealistic portrait of a mountain landscape"
|
||||
"Generate an anime-style character design"
|
||||
```
|
||||
|
||||
### Specific Details
|
||||
```
|
||||
"Generate a photo of a baby panda sitting on a rock, surrounded by bamboo, with soft lighting"
|
||||
```
|
||||
|
||||
### Variations
|
||||
```
|
||||
"Generate 3 variations of a modern logo for a coffee shop"
|
||||
```
|
||||
|
||||
## Learn More
|
||||
|
||||
- [Nano Banana Agent in Registry](/docs/guides/agent-registry#%EF%B8%8F-nano-banana-agent)
|
||||
- [Agent Configuration](/docs/guides/configuring-dexto/overview)
|
||||
- [Google Gemini Models](https://ai.google.dev/)
|
||||
19
dexto/docs/examples/intro.md
Normal file
19
dexto/docs/examples/intro.md
Normal file
@@ -0,0 +1,19 @@
|
||||
---
|
||||
sidebar_position: 1
|
||||
title: Examples
|
||||
---
|
||||
|
||||
# Examples
|
||||
|
||||
Explore practical examples showcasing Dexto's capabilities in real-world scenarios. Each example demonstrates specific features and patterns you can use in your own projects.
|
||||
|
||||
## Available Examples
|
||||
|
||||
Browse through our collection of examples to learn how to:
|
||||
- Build agents with MCP integration
|
||||
- Create interactive applications
|
||||
- Process and analyze data
|
||||
- Integrate with external services
|
||||
- Generate and manipulate media
|
||||
|
||||
Each example includes complete code and implementation details to help you get started quickly.
|
||||
87
dexto/docs/examples/mcp-integration.md
Normal file
87
dexto/docs/examples/mcp-integration.md
Normal file
@@ -0,0 +1,87 @@
|
||||
---
|
||||
title: "Adding Custom MCP Servers"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Adding Custom MCP Servers
|
||||
|
||||
Extend Dexto's capabilities by adding your own Model Context Protocol (MCP) servers with new tools and data sources.
|
||||
|
||||
<ExpandableImage src="https://github.com/user-attachments/assets/1a3ca1fd-31a0-4e1d-ba93-23e1772b1e79" alt="Add MCP Server Example" title="Adding Custom MCP Servers" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
Add custom MCP servers to:
|
||||
- Connect new tools and APIs
|
||||
- Access external data sources
|
||||
- Integrate third-party services
|
||||
- Build custom functionality
|
||||
|
||||
## How to add MCP servers
|
||||
|
||||
### Option 1: Via Web UI
|
||||
|
||||
```bash
|
||||
# Launch the Web UI
|
||||
dexto
|
||||
```
|
||||
|
||||
1. Click on "MCP Servers" in the sidebar
|
||||
2. Click "Add Server"
|
||||
3. Enter server configuration
|
||||
4. Save and the server tools become available immediately
|
||||
|
||||
### Option 2: Via agent.yml
|
||||
|
||||
Edit your agent configuration file:
|
||||
|
||||
```yaml
|
||||
# agents/my-agent.yml
|
||||
mcpServers:
|
||||
custom-server:
|
||||
type: stdio
|
||||
command: npx
|
||||
args: ['-y', 'your-mcp-server-package']
|
||||
env:
|
||||
API_KEY: $YOUR_API_KEY
|
||||
```
|
||||
|
||||
### Option 3: Via CLI
|
||||
|
||||
```bash
|
||||
# Edit agent config directly
|
||||
nano ~/.dexto/agents/your-agent.yml
|
||||
|
||||
# Or use the coding agent config
|
||||
nano ~/.dexto/agents/coding-agent/coding-agent.yml
|
||||
```
|
||||
|
||||
## Example: Adding Brave Search
|
||||
|
||||
```yaml
|
||||
mcpServers:
|
||||
web-search:
|
||||
type: stdio
|
||||
command: npx
|
||||
args: ['-y', '@modelcontextprotocol/server-brave-search']
|
||||
env:
|
||||
BRAVE_API_KEY: $BRAVE_API_KEY
|
||||
```
|
||||
|
||||
## Available MCP Servers
|
||||
|
||||
Browse 20+ ready-to-use MCP servers in the [MCP Store](/examples/mcp-store) including:
|
||||
- **Filesystem** - File operations
|
||||
- **Brave Search** - Web search
|
||||
- **GitHub** - Repository management
|
||||
- **Slack** - Team communication
|
||||
- **PostgreSQL** - Database access
|
||||
- And many more!
|
||||
|
||||
## Learn More
|
||||
|
||||
- [MCP Configuration Guide](/docs/guides/configuring-dexto/mcpConfiguration)
|
||||
- [MCP Overview](/docs/mcp/overview)
|
||||
- [MCP Manager](/docs/mcp/mcp-manager)
|
||||
- [Official MCP Servers](https://github.com/modelcontextprotocol/servers)
|
||||
68
dexto/docs/examples/mcp-store.md
Normal file
68
dexto/docs/examples/mcp-store.md
Normal file
@@ -0,0 +1,68 @@
|
||||
---
|
||||
title: "MCP Store: Tool Discovery & Integration"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# MCP Store: Tool Discovery & Integration
|
||||
|
||||
Equip your agents with 20+ MCP servers and start using them via chat - instantly.
|
||||
|
||||
<ExpandableImage src="/assets/mcp_store_demo.gif" alt="MCP Store Demo" title="MCP Store: Tool Discovery & Integration" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
The MCP Store provides a curated collection of ready-to-use MCP servers:
|
||||
- **Discover tools** from the integrated marketplace
|
||||
- **Install with one click** directly from the Web UI
|
||||
- **Bring your own keys** for API-based services
|
||||
- **Start using immediately** - no configuration needed
|
||||
|
||||
## How to use
|
||||
|
||||
1. **Launch Web UI:**
|
||||
```bash
|
||||
dexto
|
||||
```
|
||||
|
||||
2. **Open MCP Store:**
|
||||
- Click "MCP Store" in the sidebar
|
||||
- Browse available servers
|
||||
- View server details, required keys, and capabilities
|
||||
|
||||
3. **Install a server:**
|
||||
- Click "Install" on any server
|
||||
- Provide required API keys if needed
|
||||
- Server tools become available instantly
|
||||
|
||||
4. **Use in conversation:**
|
||||
```text
|
||||
"Search the web for latest AI news" # Uses Brave Search
|
||||
"List files in this directory" # Uses Filesystem tools
|
||||
"Send a message to the team channel" # Uses Slack integration
|
||||
```
|
||||
|
||||
## Available Servers
|
||||
|
||||
Browse the integrated MCP Store to discover available servers. The store includes servers across categories like:
|
||||
|
||||
- **Search & Web** - Brave Search, web scraping
|
||||
- **Development** - Filesystem access, Git operations
|
||||
- **Communication** - Slack integration
|
||||
- **Data** - Database connections, file operations
|
||||
- **AI** - Image generation, audio processing
|
||||
|
||||
Check the Web UI MCP Store to see the current list of available servers with installation instructions and requirements.
|
||||
|
||||
## Contributing
|
||||
|
||||
Can't find an MCP server you need?
|
||||
- [Contribute to the registry](https://github.com/truffle-ai/dexto/blob/main/CONTRIBUTING.md)
|
||||
- [Build your own MCP server](https://modelcontextprotocol.io/)
|
||||
- Submit a feature request
|
||||
|
||||
## Learn More
|
||||
|
||||
- [MCP Overview](/docs/mcp/overview)
|
||||
- [MCP Configuration](/docs/guides/configuring-dexto/mcpConfiguration)
|
||||
- [Official MCP Servers](https://github.com/modelcontextprotocol/servers)
|
||||
69
dexto/docs/examples/memory.md
Normal file
69
dexto/docs/examples/memory.md
Normal file
@@ -0,0 +1,69 @@
|
||||
---
|
||||
title: "Memory: Persistent Context & Learning"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Memory: Persistent Context & Learning
|
||||
|
||||
Create and save memories so your agent automatically uses them to create personalized experiences.
|
||||
|
||||
<ExpandableImage src="/assets/memory_demo.gif" alt="Memory Demo" title="Memory: Persistent Context & Learning" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
Dexto's memory system allows agents to:
|
||||
- **Remember user preferences** across sessions
|
||||
- **Learn from past interactions** to provide better responses
|
||||
- **Store important context** for future reference
|
||||
- **Personalize responses** based on saved information
|
||||
|
||||
## How it works
|
||||
|
||||
Agents automatically create and retrieve memories during conversations. You can also manually save important information:
|
||||
|
||||
```bash
|
||||
# In any Dexto session
|
||||
> "Remember that I prefer TypeScript over JavaScript"
|
||||
> "Save that my timezone is PST"
|
||||
> "Remember my favorite color is blue"
|
||||
```
|
||||
|
||||
The agent will use these memories in future conversations:
|
||||
|
||||
```bash
|
||||
# Later in a different session
|
||||
> "Create a new project for me"
|
||||
# Agent: "I'll create a TypeScript project for you since that's your preference..."
|
||||
```
|
||||
|
||||
## Memory Types
|
||||
|
||||
- **User Preferences**: Personal settings and choices
|
||||
- **Context**: Important background information
|
||||
- **Facts**: Specific details to remember
|
||||
- **Learned Patterns**: Behavioral insights from interactions
|
||||
|
||||
## Managing Memories
|
||||
|
||||
### View memories
|
||||
```bash
|
||||
dexto
|
||||
# Navigate to "Memories" in the sidebar
|
||||
```
|
||||
|
||||
### Clear memories
|
||||
Delete individual memories or clear all via the Web UI settings.
|
||||
|
||||
## Privacy
|
||||
|
||||
- Memories are stored locally by default
|
||||
- Configure storage backend (Redis, PostgreSQL, SQLite)
|
||||
- Full control over what gets saved
|
||||
- Export and import capabilities
|
||||
|
||||
## Learn More
|
||||
|
||||
- [Session Management](/docs/guides/configuring-dexto/sessions)
|
||||
- [Storage Configuration](/docs/guides/configuring-dexto/storage)
|
||||
- [Agent Configuration](/docs/guides/configuring-dexto/overview)
|
||||
94
dexto/docs/examples/playground.md
Normal file
94
dexto/docs/examples/playground.md
Normal file
@@ -0,0 +1,94 @@
|
||||
---
|
||||
title: "Playground: Interactive Development Environment"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Playground: Interactive Development Environment
|
||||
|
||||
A testing playground to view tools in your MCP servers before connecting them to LLMs to see detailed response structures.
|
||||
|
||||
<ExpandableImage src="/assets/playground_demo.gif" alt="Playground Demo" title="Playground: Interactive Development Environment" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
The Playground provides an interactive environment to:
|
||||
- **Explore MCP server tools** without running the full agent
|
||||
- **Test tool parameters** and see responses in real-time
|
||||
- **Inspect response structures** for debugging
|
||||
- **Validate server connections** before integration
|
||||
- **Develop and debug** custom MCP servers
|
||||
|
||||
## How to access
|
||||
|
||||
```bash
|
||||
# Launch Web UI
|
||||
dexto
|
||||
```
|
||||
|
||||
Navigate to "Playground" in the sidebar.
|
||||
|
||||
## Features
|
||||
|
||||
### 1. Server Browser
|
||||
- View all connected MCP servers
|
||||
- See available tools for each server
|
||||
- Inspect tool schemas and parameters
|
||||
|
||||
### 2. Tool Tester
|
||||
- Select any tool from connected servers
|
||||
- Fill in parameters with a guided form
|
||||
- Execute tools directly
|
||||
- View formatted responses
|
||||
|
||||
### 3. Response Inspector
|
||||
- See full JSON responses
|
||||
- Expand/collapse nested structures
|
||||
- Copy response data
|
||||
- View error messages and stack traces
|
||||
|
||||
### 4. Connection Validator
|
||||
- Test server connectivity
|
||||
- Verify authentication
|
||||
- Check tool availability
|
||||
- Debug connection issues
|
||||
|
||||
## Use Cases
|
||||
|
||||
### Developing MCP Servers
|
||||
Test your custom MCP server tools before integrating with agents:
|
||||
|
||||
```bash
|
||||
# Start your MCP server
|
||||
dexto --mode mcp
|
||||
|
||||
# In another terminal, test it in playground
|
||||
dexto
|
||||
```
|
||||
|
||||
### Debugging Tool Issues
|
||||
When a tool isn't working as expected, use the playground to:
|
||||
1. Verify the tool exists
|
||||
2. Check parameter requirements
|
||||
3. Test with sample inputs
|
||||
4. Inspect error responses
|
||||
|
||||
### Exploring New Servers
|
||||
Before adding a new MCP server to your agent, explore its capabilities in the playground to understand what tools it provides and how to use them.
|
||||
|
||||
## Example Workflow
|
||||
|
||||
1. **Add a server** via Web UI or config
|
||||
2. **Open Playground**
|
||||
3. **Select server** from dropdown
|
||||
4. **Choose a tool** to test
|
||||
5. **Fill parameters** using the form
|
||||
6. **Execute** and view response
|
||||
7. **Iterate** until you understand the tool behavior
|
||||
|
||||
## Learn More
|
||||
|
||||
- [MCP Overview](/docs/mcp/overview)
|
||||
- [MCP Configuration](/docs/guides/configuring-dexto/mcpConfiguration)
|
||||
- [Building MCP Servers](https://modelcontextprotocol.io/)
|
||||
- [Web UI Guide](/docs/guides/web-ui)
|
||||
45
dexto/docs/examples/podcast-agent.md
Normal file
45
dexto/docs/examples/podcast-agent.md
Normal file
@@ -0,0 +1,45 @@
|
||||
---
|
||||
title: "Podcast Agent: Generate AI Podcasts"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Podcast Agent: Generate AI Podcasts
|
||||
|
||||
Generate engaging podcast content with AI-powered audio generation featuring multiple speakers.
|
||||
|
||||
**Task:** `Generate an intro for a podcast about the latest in AI.`
|
||||
|
||||
```bash
|
||||
dexto --agent podcast-agent
|
||||
```
|
||||
|
||||
<ExpandableImage src="https://github.com/user-attachments/assets/cfd59751-3daa-4ccd-97b2-1b2862c96af1" alt="Podcast Agent Demo" title="Podcast Agent Demo" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
The Podcast Agent uses Google Gemini TTS to create multi-speaker audio content:
|
||||
- Generate podcast intros and outros
|
||||
- Create conversations between multiple hosts
|
||||
- Customize voice characteristics and speaking styles
|
||||
- Export high-quality audio files
|
||||
|
||||
## Requirements
|
||||
|
||||
- `GOOGLE_GENERATIVE_AI_API_KEY` environment variable
|
||||
- Google Gemini 2.5 Flash (included in agent config)
|
||||
|
||||
## Try it
|
||||
|
||||
```bash
|
||||
# Install the agent
|
||||
dexto install podcast-agent
|
||||
|
||||
# Run it
|
||||
dexto --agent podcast-agent
|
||||
```
|
||||
|
||||
Try prompts like:
|
||||
```
|
||||
"Generate a podcast intro with two hosts discussing the future of AI in healthcare"
|
||||
```
|
||||
64
dexto/docs/examples/portable-agents.md
Normal file
64
dexto/docs/examples/portable-agents.md
Normal file
@@ -0,0 +1,64 @@
|
||||
---
|
||||
title: "Portable Agents: Use Your Agents from Cursor"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Portable Agents: Use Your Agents from Cursor
|
||||
|
||||
Dexto agents are modular, composable, and portable - run them from anywhere including Cursor, Claude Desktop, and other MCP clients.
|
||||
|
||||
<ExpandableImage src="/img/cursor/dexto-agent-cursor.png" alt="Cursor Integration Demo" title="Portable Agents: Use Your Agents from Cursor" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
Connect to Dexto as an MCP server to use your agents from any MCP-compatible client:
|
||||
- Run Dexto agents from Cursor
|
||||
- Use agents in Claude Desktop
|
||||
- Integrate with custom MCP clients
|
||||
- Share agents across tools and environments
|
||||
|
||||
## How it works
|
||||
|
||||
1. **Start Dexto as an MCP server:**
|
||||
```bash
|
||||
dexto --mode mcp --agent podcast-agent
|
||||
```
|
||||
|
||||
2. **Configure your MCP client** (e.g., Cursor, Claude Desktop):
|
||||
```json
|
||||
{
|
||||
"mcpServers": {
|
||||
"dexto-podcast": {
|
||||
"command": "dexto",
|
||||
"args": ["--mode", "mcp", "--agent", "podcast-agent"]
|
||||
}
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
3. **Use the agent** from your MCP client just like any other tool!
|
||||
|
||||
## Example: Podcast Agent in Cursor
|
||||
|
||||
In this example, we expose the Podcast Agent as an MCP server and use it from Cursor to generate podcast intros while coding.
|
||||
|
||||
```bash
|
||||
# Start Dexto as MCP server with podcast agent
|
||||
dexto --mode mcp --agent podcast-agent
|
||||
```
|
||||
|
||||
Then in Cursor, the Podcast Agent's tools become available as native MCP tools.
|
||||
|
||||
## Benefits
|
||||
|
||||
- **Portable**: Same agent, multiple interfaces
|
||||
- **Composable**: Combine agents from different sources
|
||||
- **Consistent**: Agent behavior stays the same across clients
|
||||
- **Reusable**: Build once, use everywhere
|
||||
|
||||
## Learn More
|
||||
|
||||
- [Expose Dexto as MCP Server](/docs/mcp/dexto-as-mcp-server)
|
||||
- [Agent Configuration](/docs/guides/configuring-dexto/overview)
|
||||
- [MCP Overview](/docs/mcp/overview)
|
||||
61
dexto/docs/examples/snake-game.md
Normal file
61
dexto/docs/examples/snake-game.md
Normal file
@@ -0,0 +1,61 @@
|
||||
---
|
||||
title: "Coding Agent: Create Apps on Demand"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Coding Agent: Create Apps on Demand
|
||||
|
||||
Build full-stack applications, websites, and interactive games with AI-powered coding agents.
|
||||
|
||||
**Task:** `Can you create a snake game in a new folder and open it when done?`
|
||||
|
||||
```bash
|
||||
dexto --agent coding-agent "Can you create a snake game in a new folder and open it when done?"
|
||||
```
|
||||
|
||||
<ExpandableImage src="/assets/coding_agent_demo.gif" alt="Snake Game Development Demo" title="Coding Agent: Create Apps on Demand" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
The Coding Agent can:
|
||||
- Generate complete applications from natural language descriptions
|
||||
- Write HTML, CSS, JavaScript, TypeScript, and more
|
||||
- Create interactive games and websites
|
||||
- Automatically open finished projects in the browser
|
||||
- Refactor and debug existing code
|
||||
|
||||
## Requirements
|
||||
|
||||
- Anthropic Claude Haiku 4.5 (included in agent config)
|
||||
- Filesystem and browser tools (included)
|
||||
|
||||
## Try it
|
||||
|
||||
```bash
|
||||
# Install the agent
|
||||
dexto install coding-agent
|
||||
|
||||
# Open the agent in web UI
|
||||
dexto --agent coding-agent
|
||||
|
||||
## One shot prompts in CLI
|
||||
# Create a game
|
||||
dexto --agent coding-agent "create a snake game in HTML/CSS/JS, then open it in the browser"
|
||||
|
||||
# Build a website
|
||||
dexto --agent coding-agent "create a landing page for a coffee brand inspired by star wars"
|
||||
```
|
||||
|
||||
The agent will:
|
||||
1. Create project files and folders
|
||||
2. Write the code
|
||||
3. Open the finished app in your browser
|
||||
|
||||
## Supported Languages
|
||||
|
||||
50+ programming languages and config formats including:
|
||||
- HTML, CSS, JavaScript, TypeScript
|
||||
- Python, Go, Rust
|
||||
- React, Vue, Svelte
|
||||
- And more
|
||||
79
dexto/docs/examples/triage-agent.md
Normal file
79
dexto/docs/examples/triage-agent.md
Normal file
@@ -0,0 +1,79 @@
|
||||
---
|
||||
title: "Triage Agent: Multi-Agent Customer Support"
|
||||
---
|
||||
|
||||
import ExpandableImage from '@site/src/components/ExpandableImage';
|
||||
|
||||
# Triage Agent: Multi-Agent Customer Support
|
||||
|
||||
Create multi-agent systems that intelligently coordinate and delegate tasks among themselves based on user queries.
|
||||
|
||||
```bash
|
||||
dexto --agent triage-agent
|
||||
```
|
||||
|
||||
<ExpandableImage src="/assets/triage_agent_demo.gif" alt="Triage Agent Demo" title="Triage Agent: Multi-Agent Customer Support" width={900} />
|
||||
|
||||
## What it does
|
||||
|
||||
The Triage Agent demonstrates multi-agent collaboration:
|
||||
- **Router Agent**: Analyzes incoming requests and routes them to specialists
|
||||
- **Technical Support Agent**: Handles technical issues and troubleshooting
|
||||
- **Billing Agent**: Manages billing inquiries and account questions
|
||||
- **General Support Agent**: Handles general questions and information requests
|
||||
|
||||
## How it works
|
||||
|
||||
1. User submits a support request
|
||||
2. Triage agent analyzes the request
|
||||
3. Routes to the appropriate specialist agent
|
||||
4. Specialist agent handles the specific task
|
||||
5. Response is returned to the user
|
||||
|
||||
## Example Interactions
|
||||
|
||||
```bash
|
||||
# Technical issue
|
||||
"My API key isn't working"
|
||||
→ Routes to Technical Support Agent
|
||||
|
||||
# Billing question
|
||||
"How much does the premium plan cost?"
|
||||
→ Routes to Billing Agent
|
||||
|
||||
# General inquiry
|
||||
"What features do you offer?"
|
||||
→ Routes to General Support Agent
|
||||
```
|
||||
|
||||
## Key Features
|
||||
|
||||
- **Intelligent Routing**: Automatically determines the best agent for each request
|
||||
- **Context Preservation**: Maintains conversation context across agent handoffs
|
||||
- **Scalable**: Easy to add new specialist agents
|
||||
- **Collaborative**: Agents can consult each other when needed
|
||||
|
||||
## Try it
|
||||
|
||||
```bash
|
||||
# Install the agent
|
||||
dexto install triage-agent
|
||||
|
||||
# Run it
|
||||
dexto --agent triage-agent
|
||||
```
|
||||
|
||||
Try different types of requests:
|
||||
```
|
||||
"I have a billing question"
|
||||
"My API isn't responding"
|
||||
"What are your business hours?"
|
||||
```
|
||||
|
||||
Watch the multi-agent system communicate to get your responses.
|
||||
|
||||
## Learn More
|
||||
|
||||
- [Multi-Agent Systems Tutorial](/docs/tutorials/cli/examples/multi-agent-systems)
|
||||
- [Building a Triage System](/docs/tutorials/cli/examples/building-triage-system)
|
||||
- [Agent Configuration](/docs/guides/configuring-dexto/overview)
|
||||
Reference in New Issue
Block a user