External risk intelligence

CrewAI SSRF Vulnerability in Scraping Tool

CVE advisorySeverity: CRITICAL (CVSS 9.1)

CVE-2026-120630

CrewAI is a framework for building AI agents, often used in internal automation or developer-centric workflows. While it can be integrated into web applications, it is typically a library or agentic framework rather than a standard internet-facing gateway or edge service. Its exposure depends entirely on the specific application architecture into which the agent is embedded.

Server-Side Request Forgery

Halo Surface Signal: 3 out of 5 — possibly public-facing.

External exposure likelihood

Horizon Alert

Summary of the vulnerability and why it matters

This advisory addresses a critical vulnerability in CrewAI, an AI agent framework, stemming from how it handles URLs. An attacker could potentially exploit this by injecting malicious URLs, which the system might not properly validate, leading to unauthorized actions. The main concern is confirming if CrewAI is used within your environment and if these specific functions are exposed.

  • Flaw lets malicious URLs bypass security checks.
  • Confirms relevance and exposure if CrewAI is used.
  • Assess impact on AI agent framework usage.

Attack Path

How an attacker could exploit the issue

An attacker could exploit this vulnerability by tricking a system using CrewAI into visiting a malicious URL. This could happen if the system uses a tool like ScrapWebsiteTool and the attacker can influence the URL provided to it. Successful exploitation could allow an attacker to make the system perform unintended actions or access internal resources.

  • No special access needed.
  • Triggered by a malicious URL.
  • Risk of unintended system actions.

Live Threat

Current exploitation, exposure, and threat context

This vulnerability could allow an attacker to redirect requests made by CrewAI to arbitrary external systems. When CrewAI is configured to process URLs, such as with the ScrapWebsiteTool, an attacker could potentially inject a malicious URL. This could lead to unintended access to internal resources or other services that the CrewAI agent has network access to.

  • CrewAI agent network access.
  • Malicious URL injection.
  • Unintended access to internal resources.

Operational Fix

Recommended remediation, mitigation, and detection steps

This vulnerability in CrewAI impacts systems using its web scraping tools, potentially allowing attackers to exploit SSRF. Owners of applications and platforms integrating CrewAI should first locate all instances of the affected technology, determine their exposure and business criticality, and identify the accountable teams for remediation. Collaboration between application owners, platform engineers, and security teams will be essential for a risk-based approach to remediation.

  • Application and platform owners should take ownership.
  • Verify reachability and business criticality first.
  • Plan remediation based on identified risk.

Supplementary metadata

Validate whether this threat affects your internet-facing exposure.

Halo Threat Intelligence helps prioritize remediation with Halo Surface Signal and H/A/L/O context. Start exposure validation with a free external attack surface trial.

Frequently asked questions

What is CrewAI and why is it used?

CrewAI is a framework designed for building and orchestrating AI agents that perform automated tasks. Developers use it to create autonomous systems that can execute complex workflows, such as browsing the web or interacting with various APIs, by leveraging specialized tools like the ScrapWebsiteTool to retrieve information.

What does this SSRF vulnerability mean for CVE-2026-120630?

This CVE involves a Server-Side Request Forgery (SSRF) weakness, categorized as CWE-918. It occurs because the software's security guard and HTTP client check different, inconsistent URL patterns. This mismatch allows malicious input to slip past validation, tricking the framework into making unauthorized network requests to destinations it should not access.

How can an attacker trigger this vulnerability?

An attacker triggers this by injecting a malicious URL into a component like the ScrapWebsiteTool. If the agent accepts and processes this URL, it carries out the request on behalf of the system. Note that simply having CrewAI installed does not trigger the flaw; the system must be actively using a vulnerable tool to process externally influenced URL inputs.

Do I need to worry about this vulnerability in my environment?

According to Halo Surface Signal, risk depends on your specific architecture. CrewAI is typically a library used in internal workflows rather than an internet-facing gateway. You should assess whether your AI agents are reachable from outside networks or if they have privileged access to internal resources that could be targeted via these forged requests.

When should I take action to address this issue?

Begin by identifying all applications and services within your environment that integrate CrewAI. Once you have a complete inventory, evaluate whether those agents use URL-processing tools and assess the sensitivity of the data they can access. Coordinate with your development and security teams to prioritize risk-based remediation for any high-criticality agentic workflows.

References