Executive Summary

CVE-2026-4308 is a Server-Side Request Forgery (SSRF) vulnerability found in version 0.9.7 of the frdel/agent0ai agent-zero software, specifically in the function handle_pdf_document within the file python/helpers/document_query.py.

The vulnerability occurs because the software fetches URLs supplied by users without validating the URL scheme, hostname, or IP address. This allows an attacker to manipulate input so that the server makes HTTP requests to arbitrary internal or external endpoints.

The SSRF is not blind; the full HTTP response content is returned to the attacker, enabling them to access internal network services, cloud metadata endpoints, and sensitive data.

Useful 0 Not Useful 0

Impact Analysis

This vulnerability can have serious impacts including theft of cloud credentials from AWS, GCP, or Azure metadata services, unauthorized access to internal network services such as Redis, MongoDB, and internal APIs, network reconnaissance, port scanning, and data exfiltration from services not exposed to the internet.

In cloud deployments, it poses a critical risk by exposing IAM credentials and instance metadata. In corporate or internal networks, it allows attackers to access sensitive internal resources. In containerized environments like Docker or Kubernetes, it increases the risk of container escape or unauthorized access to internal services.

Useful 0 Not Useful 0

Compliance Impact

I don't know

Useful 0 Not Useful 0

Detection Guidance

[{'type': 'paragraph', 'content': 'This vulnerability can be detected by monitoring for unusual HTTP requests made by the agent-zero software, especially requests to internal IP ranges, cloud metadata endpoints (e.g., 169.254.169.254), or unexpected external URLs triggered by the handle_pdf_document function.'}, {'type': 'paragraph', 'content': 'One practical approach is to observe network traffic or logs for requests initiated by agent-zero to internal services such as Redis, MongoDB, or cloud metadata URLs.'}, {'type': 'paragraph', 'content': 'Since the vulnerability involves the agent making HTTP requests to attacker-controlled URLs, you can use network monitoring tools or packet capture commands like:'}, {'type': 'list_item', 'content': 'tcpdump -i <interface> host 169.254.169.254'}, {'type': 'list_item', 'content': "tcpdump -i <interface> 'dst net 10.0.0.0/8 or dst net 172.16.0.0/12 or dst net 192.168.0.0/16'"}, {'type': 'list_item', 'content': 'netstat -anp | grep agent-zero'}, {'type': 'paragraph', 'content': 'Additionally, you can check the agent-zero logs or WebUI activity for commands instructing the document_query tool to fetch documents from suspicious URLs.'}] [2, 3]

Useful 0 Not Useful 0

Mitigation Strategies

[{'type': 'paragraph', 'content': 'Immediate mitigation steps include replacing or disabling the vulnerable component since no official patches or vendor responses are available.'}, {'type': 'paragraph', 'content': 'Implement strict URL validation before the agent makes HTTP requests to prevent SSRF exploitation.'}, {'type': 'list_item', 'content': "Allow only 'http' and 'https' URL schemes."}, {'type': 'list_item', 'content': 'Resolve hostnames to IP addresses and block requests to private, loopback, link-local, and cloud metadata IP ranges (e.g., 169.254.169.254).'}, {'type': 'list_item', 'content': 'Raise exceptions or block requests that fail validation.'}, {'type': 'paragraph', 'content': 'An example Python validation function can be used to enforce these checks before calling requests.get().'}, {'type': 'paragraph', 'content': 'If possible, restrict network access of the agent-zero service to prevent it from reaching internal or cloud metadata endpoints.'}, {'type': 'paragraph', 'content': 'Monitor network traffic and logs for suspicious activity related to SSRF exploitation attempts.'}] [1, 3]

Useful 0 Not Useful 0