CVE-2026-8661: Server-Side Request Forgery in Rapid7 InsightConnect Markdown Plugin

Technical summary

CVE-2026-8661 is a Server-Side Request Forgery (SSRF) vulnerability in the markdown_to_pdf action of the Rapid7 InsightConnect Markdown Plugin on Linux. The vulnerability arises from insufficient input validation when converting Markdown to PDF. Attackers can inject HTML elements—specifically img src attributes, CSS url() functions, and @import directives—within Markdown input to trigger outbound HTTP requests from the server. The initial mitigation in version 4.0.0 disabled JavaScript execution but failed to neutralize resource-loading mechanisms that operate outside JavaScript context. The definitive fix in version 4.0.2 implements allowlist-based HTML sanitization, explicitly permitting safe tags and attributes while rejecting or neutralizing URL schemes in resource-loading contexts.

Business impact

This vulnerability enables attackers to perform reconnaissance on internal network topology, scan for open ports and services on internal infrastructure, or exfiltrate metadata from internal systems reachable from the InsightConnect instance. An attacker could also use the server as a pivot point to launch attacks against third-party systems, potentially triggering security alerts or rate-limiting that obscure the true attack origin. In environments where InsightConnect processes untrusted Markdown—such as user-submitted content, third-party integrations, or data from external APIs—the risk materializes readily. The impact is amplified if the InsightConnect instance has network access to sensitive internal systems or cloud service endpoints.

Affected systems

Rapid7 InsightConnect Markdown Plugin versions prior to 4.0.2 running on Linux are affected. The vulnerability is specific to the markdown_to_pdf action. Version 4.0.0 introduced an incomplete fix that mitigated JavaScript execution but left resource-loading vectors open. Patched versions (4.0.2 and later) eliminate the flaw through HTML allowlisting. Organizations must verify their exact plugin version and update accordingly.

Exploitability

Exploitation requires network access to an InsightConnect instance and the ability to submit Markdown input to the markdown_to_pdf action. The attack complexity is rated as high, reflecting the need for specific conditions—such as the plugin being exposed to untrusted input or an attacker having legitimate access to submit workflows. No user interaction is required once a malicious Markdown payload is processed. No publicly disclosed exploit code or active exploitation is known at this time; the vulnerability is not currently tracked in the CISA KEV catalog.

Remediation

Organizations should upgrade the Rapid7 InsightConnect Markdown Plugin to version 4.0.2 or later. This version implements comprehensive HTML sanitization via an allowlist approach, neutralizing resource-loading vectors. Before upgrading, audit workflows and integrations to identify where Markdown input originates. If the plugin processes untrusted data, apply additional input validation or network segmentation to limit the server's outbound access to only necessary destinations.

Patch guidance

Update the Rapid7 InsightConnect Markdown Plugin to version 4.0.2 or later. Verify the patch by checking the plugin version in your InsightConnect instance (typically accessible via the admin console or plugin details). Do not rely on version 4.0.0 alone; ensure the 4.0.2 release is deployed. After patching, restart affected workflows and conduct a brief functional test of markdown_to_pdf conversions to confirm no regression. Consider staging the patch in a non-production environment first if InsightConnect is business-critical.

Detection guidance

Monitor HTTP access logs on the InsightConnect server for unexpected outbound requests coinciding with markdown_to_pdf action execution. Look for requests to internal IP ranges, cloud metadata endpoints (169.254.169.254, etc.), or suspicious external domains initiated from the InsightConnect process. Inspect markdown_to_pdf action logs for errors or unusual processing times that might indicate resource-loading attempts. Network-level detection can flag outbound connections from InsightConnect to atypical destinations. Additionally, review workflow definitions for Markdown input sources and trace whether those sources are adequately controlled.

Why prioritize this

Although the CVSS score of 4.8 (MEDIUM) reflects limited direct impact—no confidentiality loss to the victim, only to the attacker's reconnaissance—the vulnerability is notable because SSRF flaws often serve as springboards for lateral movement and reconnaissance in defense-in-depth strategies. InsightConnect's role as an orchestration and automation platform means it often has broad network access. The two-year gap between the initial incomplete fix (4.0.0) and the full resolution (4.0.2) underscores the complexity of sanitization. Prioritize patching if InsightConnect processes untrusted Markdown or operates in segmented networks where it bridges trust boundaries.

Risk score, explained

The CVSS 3.1 score of 4.8 (MEDIUM) is driven by low confidentiality and integrity impact (the attacker gains indirect visibility into internal resources and can potentially interfere with internal services) balanced against high attack complexity (specific conditions must align) and no authentication requirement. The score does not account for the downstream risk of SSRF as a reconnaissance or lateral-movement vector, which depends on network architecture. In environments with strong network segmentation or no sensitive internal resources accessible from InsightConnect, the practical risk is lower. In flat networks or cloud environments where the platform has broad egress permissions, risk escalates.

Frequently asked questions