> For the complete documentation index, see [llms.txt](https://riteshs4hu.gitbook.io/infosec-notes/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://riteshs4hu.gitbook.io/infosec-notes/artificial-intelligence/ai-agents.md).

# AI Agents

### **AI Agent**

An AI Agent is an LLM **connected to tools** (scanner, exploit framework, search engine, etc.) and can take actions step by step.

* **Example**:\
  You say: *“Check if this web application is vulnerable to SQL injection.”*
  * The Agent decides:
    1. Use a web scanner or send crafted HTTP requests
    2. Analyse the responses
    3. Summarise the result (e.g., *“Login page is vulnerable; error-based injection confirmed”*).

***

### **Agentic AI**

Agentic AI is like a **security assistant that works autonomously** to achieve a bigger goal.\
It doesn’t just execute one task — it:

* Plans multiple steps
* Re-plans if something fails
* Asks clarifying questions if needed
* Works until the final objective is met.
* **Example**:\
  You say: *“Assess this web application for common vulnerabilities and prepare a report.”*\
  The Agentic AI could:
  1. Enumerate endpoints automatically.
  2. Test for issues like SQLi, XSS, weak headers, etc.
  3. Collect findings and evidence.
  4. Draft a report (with severity ratings, recommendations, and proof-of-concept payloads).\
     → All done without you guiding every single step.

***

#### Reference:

* <https://www.cloudflare.com/learning/ai/what-is-large-language-model/>
* <https://www.ibm.com/think/topics/ai-agents>