Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-109404 Overview
CVE-2026-109404 is a server-side request forgery (SSRF) vulnerability in conductor-oss conductor versions up to 3.33.0-rc10. The flaw resides in unspecified functionality of the AgentController.java file within the Agentspan Agent Runtime component. Attackers can manipulate request handling to coerce the server into issuing outbound requests to attacker-controlled destinations. Exploitation can occur remotely over the network without authentication or user interaction. The weakness is tracked under CWE-918.
Critical Impact
Unauthenticated remote attackers can abuse the Agent Runtime to issue arbitrary server-originated HTTP requests, potentially reaching internal services or cloud metadata endpoints.
Affected Products
- conductor-oss conductor versions up to and including 3.33.0-rc10
- Agentspan Agent Runtime component (AgentController.java)
- Deployments exposing the Conductor API to untrusted networks
Discovery Timeline
- 2026-10-13 - CVE-2026-109404 published to NVD
- 2026-10-13 - Last updated in NVD database
Technical Details for CVE-2026-109404
Vulnerability Analysis
The vulnerability stems from insufficient validation of user-supplied input used to construct outbound HTTP requests within the Agentspan Agent Runtime. The AgentController.java handler accepts request parameters that influence the destination of server-initiated calls. Because the destination is not restricted to an allowlist, an attacker can specify arbitrary URLs. The Conductor server then performs the request on the attacker's behalf, returning or acting upon the response.
SSRF flaws in orchestration services are particularly consequential. Conductor instances commonly run inside cloud environments with access to internal microservices, databases, and instance metadata endpoints such as 169.254.169.254. The confidentiality impact is limited per the CVSS 4.0 vector, but the ability to pivot into internal networks remains the primary risk.
Root Cause
The root cause is missing or inadequate validation of a URL or host parameter processed by the Agent Runtime. The code path constructs an outbound request using attacker-controlled input without enforcing scheme restrictions, host allowlists, or DNS rebinding protections. This maps directly to CWE-918: Server-Side Request Forgery.
Attack Vector
Exploitation requires only network access to the Conductor API endpoint handled by AgentController.java. An unauthenticated attacker submits a crafted request specifying a target URL. The Conductor server then issues an HTTP request to the specified destination from its own network position. Attackers typically target internal-only services, cloud metadata APIs, or non-HTTP services reachable via protocol smuggling. See the VulDB entry for CVE-2026-109404 for additional technical references.
Detection Methods for CVE-2026-109404
Indicators of Compromise
- Outbound HTTP requests from Conductor server processes to internal RFC1918 addresses or 169.254.169.254 not matching expected agent workflows
- Unusual DNS lookups originating from the Conductor host for internal-only hostnames
- Access log entries in AgentController endpoints containing URL parameters pointing to localhost, loopback, or metadata ranges
Detection Strategies
- Inspect application logs for requests to Agent Runtime endpoints that include user-controlled URL fields and correlate with subsequent outbound connections
- Deploy egress network monitoring to flag Conductor-originated traffic that deviates from documented destinations
- Baseline normal agent task destinations and alert on connections to private IP ranges, link-local addresses, or non-HTTP ports
Monitoring Recommendations
- Enable verbose request logging on Conductor API gateways and forward to a centralized SIEM for retention and correlation
- Capture DNS query telemetry from Conductor hosts and alert on resolutions to internal zones from unexpected processes
- Monitor cloud metadata service access logs where available (for example, IMDSv2 token issuance patterns) for anomalous requests originating from Conductor workloads
How to Mitigate CVE-2026-109404
Immediate Actions Required
- Restrict network access to the Conductor API so that only trusted clients can reach AgentController endpoints
- Enforce egress filtering on Conductor hosts to block connections to private IP ranges and cloud metadata endpoints unless explicitly required
- Require IMDSv2 on AWS workloads to limit SSRF-based credential theft via 169.254.169.254
- Audit recent Conductor access logs for suspicious URL parameters indicating exploitation attempts
Patch Information
At the time of publication, no fixed version has been identified in the available advisories. Monitor the conductor-oss project and the VulDB vulnerability record #416946 for a patched release. Upgrade to a version later than 3.33.0-rc10 once a fix is published by the maintainers.
Workarounds
- Place Conductor behind an authenticating reverse proxy and deny unauthenticated access to Agent Runtime routes
- Deploy a forward proxy that enforces a strict allowlist of destinations the Conductor server is permitted to reach
- Run Conductor in a network segment that has no route to internal management interfaces or cloud metadata services
- Disable or remove the Agentspan Agent Runtime component if it is not required for your workflows
# Example egress restriction using iptables to block cloud metadata access
iptables -A OUTPUT -m owner --uid-owner conductor -d 169.254.169.254 -j DROP
iptables -A OUTPUT -m owner --uid-owner conductor -d 10.0.0.0/8 -j DROP
iptables -A OUTPUT -m owner --uid-owner conductor -d 172.16.0.0/12 -j DROP
iptables -A OUTPUT -m owner --uid-owner conductor -d 192.168.0.0/16 -j DROP
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

