amit/quantum-ai-v3
1
0
Fork 0
mirror of https://github.com/thecyberlearn/quantum-ai-v3.git synced 2026-08-18 16:12:57 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

📚 Modernize agent creation documentation - remove complexity and outdated patterns

Browse Source 
**Complete Rewrite:**
✅ **Simplified Process** - 3 steps: JSON file → git commit → done
✅ **Current Examples** - Clean, working examples for both agent types
✅ **Modern Architecture** - References new modular view structure
✅ **Current Agent List** - Updated with all 8 working agents
✅ **Focused Content** - Removed outdated complexity and old patterns

**Removed:**
❌ Old database commands and complex setup procedures
❌ Outdated populate_agents references and troubleshooting
❌ Complex template creation steps that are no longer needed
❌ References to old view structure patterns

**Added:**
✅ Clean JSON examples that actually work
✅ Current category list with descriptions
✅ New modular view architecture guidance
✅ Railway auto-deployment process
✅ Simple 3-step creation workflow

**Result:** Documentation now matches the elegant simplicity of the current system

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
Claude 2025-08-14 22:59:03 +05:30
parent 8fc9eed165
commit 67ab31f550
1 changed files with 106 additions and 220 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

316
docs/AGENT_CREATION.md
View File

@ -1,72 +1,47 @@
# Agent Creation Guide # Agent Creation Guide
This guide provides comprehensive instructions for adding new agents to the Quantum Tasks AI platform. Modern guide for adding new agents to the Quantum Tasks AI platform using the streamlined file-based system.
## Overview ## Overview
Quantum Tasks AI supports **TWO DISTINCT AGENT SYSTEMS**: Quantum Tasks AI uses a **simple file-based agent system**:
1. Create JSON configuration file
2. Commit to git
3. Agent appears in marketplace automatically
- **Webhook Agents** - N8N integrations for complex processing with dynamic forms **Two Agent Types:**
- **Direct Access Agents** - External form services (JotForm, Google Forms) with embedded interfaces - **Webhook Agents** - N8N integrations with dynamic forms
- **Direct Access Agents** - External forms (JotForm, etc.) with payment processing
**⚡ RECOMMENDED APPROACH:** Use JSON configuration + `populate_agents` command for error-free, Railway-ready agent creation. ## Current Agents (8 Total)
--- ### Webhook Agents (4)
- **Social Ads Generator** - 6.00 AED - Social media ad creation
- **Job Posting Generator** - 10.00 AED - Professional job postings
- **PDF Summarizer** - 8.00 AED - Document analysis and summarization
- **5 Whys Analyzer** - 15.00 AED - Interactive root cause analysis
## Current Agent Status ### Direct Access Agents (4)
- **CyberSec Career Navigator** - FREE - Career guidance
- **AI Brand Strategist** - FREE - Brand strategy consultation
- **Lean Six Sigma Expert** - FREE - Process improvement
- **SWOT Analysis Expert** - FREE - Strategic analysis
**Total Agents: 8** (4 webhook + 4 direct access) ## Categories (6 Available)
**Total Categories: 6** Use existing categories first to avoid proliferation:
### Webhook Agents (N8N Integration) - **`analysis`** 🧠 - Problem-solving, strategic analysis
1. **Social Ads Generator** - 6.00 AED - Creates social media advertisements - **`career-education`** 🎓 - Career guidance, professional development
2. **Job Posting Generator** - 10.00 AED - Creates professional job postings - **`document-processing`** 📄 - PDF analysis, file processing
3. **PDF Summarizer** - 8.00 AED - Analyzes and summarizes PDF documents - **`human-resources`** 💼 - Job postings, HR automation
4. **5 Whys Analyzer** - 15.00 AED - Interactive chat-based root cause analysis - **`marketing`** 📢 - Social ads, content marketing
- **`consulting`** 💼 - Business consultation, expert advice
### Direct Access Agents (External Forms) ## Creating New Agents
1. **CyberSec Career Navigator** - FREE - Career guidance consultation
2. **AI Brand Strategist** - FREE - Brand strategy consultation
3. **Lean Six Sigma Expert** - FREE - Process improvement consultation
4. **SWOT Analysis Expert** - FREE - Strategic business analysis
---
## 🏷️ Choose Existing Category First
**IMPORTANT:** Always use existing categories before creating new ones to avoid category proliferation.
### Available Categories:
- 🧠 **`analysis`** - Problem-solving, SWOT analysis, strategic analysis tools
- 🎓 **`career-education`** - Career guidance, educational resources, professional development
- 📄 **`document-processing`** - PDF analysis, file processing, document tools
- 💼 **`human-resources`** - Job postings, HR automation, talent management
- 📢 **`marketing`** - Social ads, branding, content marketing, advertising
- 💼 **`consulting`** - Business consultation, strategy services, expert advice
**Only create new categories when absolutely necessary and logically distinct.**
---
## 🚀 Agent Creation Workflow
### System 1: Webhook Agents (N8N Integration)
- **Use for:** Dynamic forms, server-side processing, file uploads, complex workflows
- **Examples:** Social Ads Generator, PDF Summarizer, Job Posting Generator
- **Flow:** Marketplace → Agent detail page → Dynamic form → N8N webhook → Results
### System 2: Direct Access Agents (External Forms)
- **Use for:** External form services (JotForm, Google Forms), consultation interfaces
- **Examples:** SWOT Analysis Expert, CyberSec Career Navigator, AI Brand Strategist
- **Flow:** Marketplace → Payment processing → Quantum Tasks header + embedded external form
---
## 📝 Implementation Steps
### Step 1: Create JSON Configuration ### Step 1: Create JSON Configuration
Create a new file in `agents/configs/agents/your-agent-name.json`: Add new file in `agents/configs/agents/your-agent-name.json`:
#### Webhook Agent Example: #### Webhook Agent Example:
```json ```json
@ -74,7 +49,7 @@ Create a new file in `agents/configs/agents/your-agent-name.json`:
"slug": "content-optimizer", "slug": "content-optimizer",
"name": "Content Optimizer", "name": "Content Optimizer",
"short_description": "AI-powered content optimization and enhancement", "short_description": "AI-powered content optimization and enhancement",
"description": "Enhance your content for better engagement with AI-powered optimization suggestions, tone analysis, and improvement recommendations.", "description": "Enhance your content with AI-powered optimization suggestions, tone analysis, and improvement recommendations.",
"category": "marketing", "category": "marketing",
"price": 5.0, "price": 5.0,
"agent_type": "form", "agent_type": "form",
@ -109,97 +84,59 @@ Create a new file in `agents/configs/agents/your-agent-name.json`:
#### Direct Access Agent Example: #### Direct Access Agent Example:
```json ```json
{ {
"slug": "business-strategist", "slug": "business-coach",
"name": "Business Strategist", "name": "Business Coach",
"short_description": "Expert business strategy consultation", "short_description": "Expert business coaching consultation",
"description": "Get professional business strategy insights and recommendations from experienced consultants to grow your business effectively.", "description": "Get professional business coaching insights and strategic guidance from experienced consultants.",
"category": "consulting", "category": "consulting",
"price": 0.0, "price": 0.0,
"agent_type": "form", "agent_type": "form",
"system_type": "direct_access", "system_type": "direct_access",
"form_schema": { "form_schema": {"fields": []},
"fields": [] "webhook_url": "https://form.jotform.com/your-form-id",
},
"webhook_url": "https://agent.jotform.com/your-form-id",
"access_url_name": "agents:direct_access_handler", "access_url_name": "agents:direct_access_handler",
"display_url_name": "agents:direct_access_display" "display_url_name": "agents:direct_access_display"
} }
``` ```
### Step 2: Run populate_agents Command ### Step 2: Commit to Git
```bash ```bash
# Development git add agents/configs/agents/your-agent-name.json
source venv/bin/activate git commit -m "Add new agent: Your Agent Name"
python manage.py populate_agents git push
# Production (Railway)
python manage.py populate_agents # Runs automatically on deployment
``` ```
### Step 3: Additional Setup (Direct Access Agents Only) ### Step 3: Done!
- **Development:** Restart server to see new agent
- **Production:** Railway auto-deploys and agent appears in marketplace
For direct access agents that need custom templates or marketplace integration: ## JSON Configuration Reference
#### 3a. Create Custom Template (optional): ### Required Fields:
```html - `slug` - URL identifier (kebab-case)
<!-- templates/your_agent_name.html --> - `name` - Display name
{% extends 'base.html' %} - `short_description` - Brief marketplace description
{% load static %} - `description` - Full description
- `category` - Must match existing category
- `price` - Price in AED (0.0 for free)
- `agent_type` - Always "form"
- `system_type` - "webhook" or "direct_access"
{% block title %}Your Agent Name - Quantum Tasks AI{% endblock %} ### System-Specific Fields:
{% block extra_css %} **Webhook Agents:**
<style> - `form_schema` - Form field definitions
.main-container { max-width: none; padding: 0; height: calc(100vh - 80px); } - `webhook_url` - N8N webhook endpoint
.iframe-container { width: 100%; height: 100%; } - `access_url_name` - Empty ""
.iframe-container iframe { width: 100%; height: 100%; border: none; display: block; } - `display_url_name` - Empty ""
.footer { display: none !important; }
</style>
{% endblock %}
{% block content %} **Direct Access Agents:**
<div class="iframe-container"> - `form_schema` - Usually `{"fields": []}`
<iframe src="{{ form_url }}" frameborder="0" scrolling="auto" title="Your Agent Name"></iframe> - `webhook_url` - External form URL
</div> - `access_url_name` - "agents:direct_access_handler"
{% endblock %} - `display_url_name` - "agents:direct_access_display"
```
#### 3b. Add Custom Views (if needed): ## Form Field Types (Webhook Agents)
Add view functions to `agents/views.py` following the pattern of existing direct access agents.
#### 3c. Add URL Routes (if needed):
Add routes to `agents/urls.py` following the pattern of existing direct access agents.
#### 3d. Update Marketplace Template (if needed):
Add button logic to `agents/templates/agents/marketplace.html` for custom marketplace behavior.
**⚠️ Important**: Keep all "Try Now" buttons consistent with the format `Try Now →` (no icons or emojis).
### Step 4: Setup External Services
#### For Webhook Agents:
- Create N8N workflow at the webhook URL
- Configure webhook to accept JSON payload with `sessionId`, `message`, etc.
#### For Direct Access Agents:
- Create external form (JotForm, Google Forms, etc.)
- Ensure form URL is accessible and properly configured
---
## ✅ Benefits of This Approach
- ✅ **Single source of truth** - JSON configs define everything
- ✅ **Railway-ready immediately** - No manual database setup needed
- ✅ **Error-free** - No category creation mistakes or typos
- ✅ **Consistent** - All agents use same reliable creation process
- ✅ **Scalable** - Easy to add 100+ agents
- ✅ **Version controlled** - Configs are tracked in git
---
## 🔧 Supported Form Field Types (Webhook Agents)
- `text` - Single-line text input - `text` - Single-line text input
- `textarea` - Multi-line text input - `textarea` - Multi-line text input
@ -208,95 +145,44 @@ Add button logic to `agents/templates/agents/marketplace.html` for custom market
- `url` - URL input with validation - `url` - URL input with validation
- `checkbox` - Boolean checkbox - `checkbox` - Boolean checkbox
--- ## Custom Integration (Advanced)
## ⚠️ Common Mistakes to Avoid For agents needing custom behavior, add views to appropriate modules:
1. **Creating unnecessary categories** - Use existing ones first - **API endpoints:** `agents/api_views.py`
2. **Missing system_type** - Include "webhook" or "direct_access" - **Chat functionality:** `agents/chat_views.py`
3. **Wrong access_url_name** - Empty for webhook agents, populated for direct access - **Web interfaces:** `agents/web_views.py`
4. **Forgetting populate_agents** - Run after creating JSON config - **Direct access handlers:** `agents/direct_access_views.py`
5. **Complex custom commands** - Use JSON + populate_agents instead - **Utilities:** `agents/utils.py`
6. **Adding icons to Try Now buttons** - Keep all marketplace buttons consistent with "Try Now →" format
Then add URL routes in `agents/urls.py` and update marketplace template if needed.
## External Services
### For Webhook Agents:
- Create N8N workflow at webhook URL
- Configure to accept JSON payload with `sessionId`, `message`, etc.
### For Direct Access Agents:
- Create external form (JotForm, Google Forms, etc.)
- Ensure form URL is publicly accessible
## Railway Deployment
**Automatic Process:**
1. ✅ Git push triggers Railway deployment
2. ✅ Agent files are processed automatically
3. ✅ New agents appear in production marketplace
4. ✅ No manual database work required
## Quick Tips
- **Use existing categories** - Avoid creating unnecessary new categories
- **Keep descriptions clear** - Users should understand what the agent does
- **Test webhook URLs** - Ensure N8N endpoints are accessible
- **Free vs Paid** - Set price to 0.0 for free agents
- **Consistent naming** - Use kebab-case for slugs, Title Case for names
--- ---
## 🔄 Agent Management Commands *Last updated: 2025-01-14*
### Essential Commands:
```bash
# Populate all agents from JSON configs (main command)
python manage.py populate_agents
# Clean up expired chat sessions
python manage.py cleanup_expired_sessions
```
### Development Workflow:
1. Create JSON config file
2. Run `populate_agents`
3. Test agent functionality
4. Commit changes to git
5. Deploy to Railway (auto-runs populate_agents)
---
## 📊 Agent Configuration Reference
### Required JSON Fields:
- `slug` - URL-friendly identifier (kebab-case)
- `name` - Display name
- `short_description` - Brief description for marketplace
- `description` - Full description with details
- `category` - Must match existing category slug
- `price` - Price in AED (0.0 for free agents)
- `agent_type` - Always "form"
- `system_type` - "webhook" or "direct_access"
### System-Specific Fields:
#### Webhook Agents:
- `form_schema` - JSON schema defining form fields
- `webhook_url` - N8N webhook endpoint
- `access_url_name` - Empty string ""
- `display_url_name` - Empty string ""
#### Direct Access Agents:
- `form_schema` - Usually `{"fields": []}`
- `webhook_url` - External form URL (JotForm, etc.)
- `access_url_name` - "agents:direct_access_handler"
- `display_url_name` - "agents:direct_access_display"
---
## 🚀 Railway Deployment
When you commit changes to the repository:
1. ✅ **Railway auto-deploys** new code
2. ✅ **populate_agents runs automatically** on deployment
3. ✅ **New agents appear** in production marketplace
4. ✅ **Categories are created** if needed (but use existing ones first!)
5. ✅ **No manual database work** required
---
## 📞 Support & Documentation
- **Main Documentation**: See `CLAUDE.md` for project overview
- **Agent Issues**: Check Railway logs and database for agent status
- **Form Problems**: Verify external form URLs are accessible
- **Category Issues**: Use existing categories from the list above
---
## 🚀 Quick Agent Request
**For fast agent creation**, use the **Agent Request Template**:
👉 **See `docs/AGENT_REQUEST_TEMPLATE.md`** for a simple template to request new agents
Simply fill out the template and provide it to Claude Code for instant agent creation!
---
*Last updated: 2025-01-08*