CVE-2026-93982 Overview
OpenPanel through commit bad75bdd writes Model Context Protocol (MCP) authentication tokens from URL query parameters to plaintext application logs without redaction. The flaw is categorized under [CWE-532] (Insertion of Sensitive Information into Log File) and affects OpenPanel deployments that expose the MCP endpoint. Attackers with access to application stdout, container logs, or centralized logging systems can capture base64-encoded credentials and replay MCP requests to access project analytics data.
Critical Impact
Log-based token exposure enables credential replay against the MCP interface, allowing unauthorized access to OpenPanel project analytics without triggering authentication anomalies.
Affected Products
- OpenPanel through commit bad75bdd
- OpenPanel deployments exposing the Model Context Protocol endpoint
- Environments forwarding OpenPanel stdout to centralized logging systems
Discovery Timeline
- 2026-09-19 - CVE-2026-93982 published to NVD
- 2026-09-22 - Last updated in NVD database
Technical Details for CVE-2026-93982
Vulnerability Analysis
OpenPanel accepts MCP authentication tokens through URL query parameters on incoming HTTP requests. The application logs the full request URL to stdout during normal request processing. Because no redaction layer strips sensitive query parameters before serialization, the base64-encoded token is written verbatim to application logs.
Any principal that can read those logs, including operators, log aggregation pipelines, sidecar containers, or downstream SIEM ingestion accounts, gains access to a working credential. The token remains valid until rotated, and replay does not require the original client context.
Root Cause
The root cause is missing sanitization of sensitive URL query parameters before writing request metadata to application logs. Passing bearer-style credentials through query strings is discouraged precisely because URLs are logged by default across web frameworks, reverse proxies, and observability stacks. OpenPanel did not implement a redaction filter for the token query parameter on the MCP handler path.
Attack Vector
Exploitation requires local read access to the log stream or downstream storage. The attacker does not need to interact with the OpenPanel API directly during collection. After harvesting a token, the attacker issues authenticated MCP requests from any network location the endpoint accepts, retrieving project analytics data. The attack does not require user interaction or elevated privileges beyond log read access.
See the GitHub Security Advisory GHSA-8wx6-g25r-2943 and the VulnCheck Advisory on OpenPanel MCP for the vendor-published technical description.
Detection Methods for CVE-2026-93982
Indicators of Compromise
- Log entries containing MCP endpoint URIs with a populated token= query parameter
- MCP API requests originating from IP addresses not associated with the legitimate client fleet
- Repeated successful MCP authentications using the same token from disparate source networks
- Access to log files or logging system indexes by accounts outside the platform operations team
Detection Strategies
- Grep application log archives for patterns matching mcp request paths combined with token= query values, then treat matches as exposed credentials
- Correlate MCP access logs against a baseline of authorized client identifiers and flag deviations
- Alert on read operations against logging indexes by non-operations service accounts
- Track token reuse across geographically or organizationally distinct source addresses
Monitoring Recommendations
- Forward OpenPanel logs to a SIEM with field-level redaction applied at ingestion
- Enable audit logging on centralized logging platforms to record who queries indexes containing OpenPanel data
- Monitor MCP endpoint access rates and response volumes for anomalies indicative of analytics scraping
- Review CI/CD and container platform logs for OpenPanel URLs exposed in build output or debug traces
How to Mitigate CVE-2026-93982
Immediate Actions Required
- Rotate all MCP authentication tokens issued before applying the fix, treating prior tokens as compromised
- Restrict read access to OpenPanel application logs and any downstream aggregation systems to a minimal operator set
- Purge or redact historical log archives that contain MCP URLs with token query parameters
- Move MCP clients off query-parameter authentication where the deployment supports header-based token submission
Patch Information
Refer to the GitHub Security Advisory GHSA-8wx6-g25r-2943 for the fixed commit and remediation guidance. Deployments running any build up to and including commit bad75bdd are affected and should upgrade to a release that includes the redaction fix.
Workarounds
- Deploy a reverse proxy in front of OpenPanel that strips the token query parameter from access logs before forwarding
- Configure the logging framework to apply a regex-based redaction filter to inbound request URIs
- Route MCP traffic over an authenticated tunnel and disable query-parameter token acceptance at the ingress layer
- Shorten token lifetimes and enforce automated rotation to limit replay windows if exposure recurs
# Example nginx redaction of the token query parameter before logging
log_format mcp_safe '$remote_addr - $remote_user [$time_local] '
'"$request_method $uri" $status $body_bytes_sent';
server {
location /mcp/ {
access_log /var/log/nginx/openpanel_mcp.log mcp_safe;
proxy_pass http://openpanel_upstream;
}
}
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

