Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessAuthenticate with low-privilege account
-
DeliveryCreate or access agent with Invoke component
-
ExploitSupply internal or metadata destination URL
-
ExecutionServer issues HTTP request without safety validation
-
ImpactRetrieve internal service response or cloud IAM credentials
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A01:2025 - Broken Access Control
ASVS 5.0
V4 - API and Web Service URL allowlisting, SSRF prevention
MITRE ATT&CK
2 techniques across 2 tactics: Initial Access, Credential Access The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Exploitation requires a valid account on the RAGFlow instance with permission to create, modify, or trigger agent workflows - a low-privilege but authenticated condition (PR:L per CVSS). For the highest-impact path targeting cloud instance metadata endpoints, no special non-default product configuration is required beyond the use of the Invoke component in an agent workflow, which is a standard product capability. When an agent is configured to interpolate chat query content into the Invoke URL, the effective attacker pool expands to any user authorized to send chat queries to that agent, not just users with workflow editing rights. Exploitation is constrained by the network reachability of internal target destinations from the RAGFlow server process, and by IMDSv2 enforcement on cloud platforms - AWS EC2 IMDSv2 requires a PUT-based token exchange that blocks naive single-hop SSRF against the metadata endpoint, raising the practical barrier on hardened cloud deployments. |
| Impact | The CVSS 4.0 score of 8.4 reflects a network-accessible, low-complexity exploit requiring only low-privilege authentication (PR:L), with high confidentiality impact on both the vulnerable system (VC:H) and subsequent systems (SC:H), accurately capturing the risk of cloud credential theft via instance metadata endpoints and enumeration of co-located internal services. No EPSS data was supplied and the vulnerability does not appear in the CISA KEV catalog, so mass exploitation has not been confirmed by any authoritative source. However, two public patch commits on GitHub clearly identify the missing validation call, making the exploit mechanism straightforward to reconstruct - even a mid-sized deployment team in walbrzych running RAGFlow on a cloud-hosted VM should treat this as urgent, given the near-zero complexity required to reach AWS or GCP metadata endpoints once a low-privilege account is available. The gap between low-privilege authenticated access (PR:L) and high downstream confidentiality impact on co-located internal infrastructure (SC:H) makes this a genuine operational priority rather than a theoretical high-score edge case. |
| Exploit ScenarioDerived | An authenticated low-privilege RAGFlow user accesses an agent whose Invoke component URL is configured to interpolate user-supplied chat query content into the destination field. The user sends a chat query containing an internal target such as http://169.254.169.254/latest/meta-data/iam/security-credentials/ and the RAGFlow server issues the HTTP request without validation, returning the cloud IAM role's temporary credentials in the agent output. Two public patch commits on GitHub clearly identify the missing assert_url_is_safe call, making the vulnerable code path straightforward to reconstruct for any reviewer examining the diff. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Aug 18, 2026
Server-side request forgery in RAGFlow before v0.26.3 allows authenticated users who can create or trigger agent workflows to direct the server to make HTTP requests to arbitrary internal destinations, including RFC 1918 addresses, loopback interfaces, link-local ranges, and cloud instance metadata endpoints. The vulnerable 'Invoke' component in agent/component/invoke.py bypasses the shared assert_url_is_safe validator consistently applied in other fetch paths (crawler, SearXNG, file-upload, RSS), and returns the fetched response body directly to the caller. No public exploit has been identified at time of analysis, though two patch commits are publicly visible on GitHub and the vulnerable code location is documented in the references.
RemediationAI
Upgrade to RAGFlow v0.26.3 or later, released July 2, 2026, which addresses this vulnerability via commits e16d1a0150e1ca069beb538ae3dcd03f59edc5fa and c4fe68eaa0bf1d6442d2cd6ac2e35bc9ccbed34f (see https://github.com/infiniflow/ragflow/releases/tag/v0.26.3). If immediate upgrade is not feasible, restrict the ability to create or modify agent workflows to highly trusted administrative users only, which eliminates the direct URL configuration path. Additionally, disable chat query interpolation into Invoke component URL fields if that pattern is in use, as it exposes the SSRF to any user with chat access to the affected agent. Network-level egress filtering on the RAGFlow host to block RFC 1918 ranges (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16), loopback (127.0.0.0/8), and link-local (169.254.0.0/16) addresses will reduce the blast radius but does not fix the underlying vulnerability and may disrupt legitimate outbound integrations that target internal services. Cloud deployments should enforce IMDSv2 on AWS EC2 instances, requiring a PUT-based token exchange that blocks single-hop SSRF against the metadata endpoint.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
RAGFlow (CPE: cpe:2.3:a:infiniflow:ragflow:*:*:*:*:*:*:*:*) is an open-source Retrieval-Augmented Generation framework developed by InfiniFlow. The 'Invoke' agent component (agent/component/invoke.py) provides workflow pipelines with the ability to make outbound HTTP requests to configurable URLs constructed from canvas configuration and runtime template variables. The root cause is CWE-918 (Server-Side Request Forgery): the component passes the resolved URL directly to Python's requests.get, requests.post, or requests.put without invoking the assert_url_is_safe validator or pinning the resolved IP address to prevent DNS rebinding - security controls that are present in the crawler, SearXNG, file-upload, and RSS fetch paths. When an agent is configured to interpolate chat query content into the Invoke URL field, the effective attacker population expands from users with agent-editing rights to any user authorized to send chat queries to that agent.
Know when Ragflow gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:L/VA:N/SC:H/SI:L/SA:N
Network-accessible SSRF exploitable by any authenticated user (PR:L); scope change (S:C) reflects high confidentiality impact on reachable internal and cloud metadata systems.
Primary rating from Vendor (VulnCheck).
CVSS VectorVendor: VulnCheck
DescriptionCVE.org
RAGFlow before 0.26.3 contains a server-side request forgery vulnerability in the agent workflow "Invoke" component (agent/component/invoke.py). The component builds an outbound request URL from canvas configuration and runtime template variables and passes it to requests.get, requests.post, or requests.put without calling the shared assert_url_is_safe validator or pinning the resolved address, unlike the crawler, SearXNG, file-upload, and RSS fetch paths. A user who can create or trigger an agent can direct the server to fetch loopback, link-local, and RFC 1918 destinations, including cloud instance metadata endpoints and services co-located on the deployment network, and the response body is returned as the component output. Where an agent is configured to interpolate the chat query into the Invoke URL, the destination is chosen by whoever can send that query.
Executive ViewAI
RAGFlow versions before 0.26.3 contain a Server-Side Request Forgery vulnerability affecting authenticated users with agent workflow creation privileges, enabling them to direct servers to request internal systems, cloud metadata endpoints, and RFC 1918 addresses. Attackers could extract sensitive data, credentials, and cloud configuration details from environments where RAGFlow has network access. Organizations running RAGFlow must upgrade immediately and audit access to the workflow creation capability.
Apply vendor patch.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Server-Side Request Forgery exploits applications that fetch remote resources based on user-supplied URLs.
A vulnerability in infiniflow/ragflow versions v0.12.0 allows for remote code execution. Rated critical severity (CVSS 9
Path traversal vulnerability in RAGFlow RAG engine version 0.23.1 allows unauthenticated attackers to read arbitrary fil
In infiniflow/ragflow versions 0.12.0, the `web_crawl` function in `document_app.py` contains multiple vulnerabilities.
RAGFlow is an open-source RAG (Retrieval-Augmented Generation) engine. Rated high severity (CVSS 8.9), this vulnerabilit
RAGFlow versions prior to 0.22.0 generate API keys and beta assistant/agent share tokens with a predictable scheme, lett
The `add_llm` function in `llm_app.py` in infiniflow/ragflow version 0.11.0 contains a remote code execution (RCE) vulne
Remote code execution in RAGFlow versions prior to 0.23.0 allows a low-privileged authenticated user (a normal login acc
RAGFlow is an open-source RAG (Retrieval-Augmented Generation) engine based on deep document understanding. Rated high s
A Server-Side Request Forgery (SSRF) vulnerability exists in infiniflow/ragflow version 0.12.0. Rated high severity (CVS
RAGFlow 0.13.0 suffers from improper access control in document-hooks.ts, allowing unauthorized access to user documents
A vulnerability in infiniflow/ragflow version RAGFlow-0.13.0 allows for partial account takeover via insecure data query
Server-side template injection in RAGFlow 0.24.0 and earlier allows any authenticated user to execute arbitrary operatin
EUVD-2026-60824
GHSA-9m4v-54qq-67fw
NVD
CVE.org
EPSS
Vulnrichment
Exploit-DB