Your First Task
Learn how to run your first AI-assisted security task with Memfit AI.
Starting Memfit AI
Via Yakit
- Launch Yakit
- Navigate to the AI Agent panel
- Ensure the Yaklang engine is connected
Via Command Line
yak ai
Simple Task: Port Scanning
Let's start with a simple task to understand how Memfit AI works.
Enter Your Request
In the chat interface, type:
Scan the ports of localhost and tell me what services are running
Observe the ReAct Loop
Memfit AI will enter Instant Execution Mode because this is a clear, atomic task:
[Thought] User wants to scan localhost ports. I'll use a port scanner.
[Action] PortScan
target: localhost
ports: 1-1000
[Observation]
Port 22: SSH
Port 80: HTTP
Port 443: HTTPS
[Thought] Scan complete. I'll summarize the findings.
[Response] Found 3 open ports on localhost:
- Port 22: SSH service
- Port 80: HTTP web server
- Port 443: HTTPS secure web server
Complex Task: Security Audit
Now let's try a more complex task that triggers Plan-Execute Mode.
Enter Your Request
Perform a security assessment of the web application at http://localhost:8080
Observe Plan Generation
Memfit AI generates a structured plan:
Security Assessment Plan:
1. Reconnaissance
├── Technology fingerprinting
└── Directory enumeration
2. Vulnerability Scanning
├── Common vulnerability checks
└── Configuration analysis
3. Analysis & Reporting
├── Finding prioritization
└── Report generation
Would you like to proceed with this plan?
Review and Approve
You can:
- Approve - Proceed with the plan as-is
- Modify - Adjust the plan before execution
- Cancel - Stop the operation
Watch Execution
After approval, multiple ReAct loops execute the plan:
[Executing: Reconnaissance]
→ Technology fingerprinting... Done
→ Found: nginx, PHP, MySQL
[Executing: Vulnerability Scanning]
→ Common vulnerability checks... Done
→ Found: 2 potential issues
[Generating Report...]
Understanding the Output
Execution Trace
Each step shows:
- Thought - AI reasoning
- Action - Tool being used
- Observation - Results received
Final Report
Comprehensive findings with:
- Severity levels
- Detailed descriptions
- Remediation recommendations
Key Concepts Demonstrated
| Concept | Example |
|---|---|
| ReAct Loop | Port scanning thought-action-observation cycle |
| Plan-Execute | Security assessment task decomposition |
| Human-in-the-Loop | Plan review before execution |
| Tool Usage | PortScan, WebScan tools |
Next Steps
- Basic Concepts - Deeper understanding
- Configuration - Customize behavior
- Tool Reference - Available tools