DEV Community

facilishz
facilishz

Posted on

AutoSys Workload Automation's November fixes include eleven privilege and credential paths

AutoSys Workload Automation's November fixes include eleven privilege and credential paths

AutoSys Workload Automation is a scheduling and job-control plane. Its Workload Control Center (WCC) can issue work to managed hosts, its engine coordinates jobs, and its agents execute them. That design makes a missing authorization check more consequential than a similar bug in a standalone dashboard. Broadcom's November 17 notice, CA20261117-01, rates the set of vulnerabilities critical with a CVSS base score of 9.3 and lists eleven CVEs.

The notice identifies problems that cross authentication, logging, and execution boundaries:

  • CVE-2026-126547 is a missing authorization check in a WCC web interface that could let an unauthenticated remote attacker invoke agent control operations.
  • CVE-2026-126548 is an authenticated server-side request forgery in a WCC export feature.
  • CVE-2026-126549 can write recoverable credentials from failed authentication attempts to an application log.
  • CVE-2026-126550 can write active session identifiers to logs, including administrator sessions.
  • CVE-2026-126551 involves a hard-coded cryptographic key in WCC and could allow remote code execution on managed System Agent hosts.
  • CVE-2026-126552 affected older releases where an unauthenticated WCC configuration endpoint could execute operating-system commands.
  • CVE-2026-126553 affected older releases where an endpoint deserialized untrusted data and could execute commands on the WCC server.

The remaining identifiers in the notice are CVE-2026-126555, CVE-2026-126556, CVE-2026-126557, and CVE-2026-126558. Broadcom lists them as part of the same critical advisory, but the public notice excerpt does not give each one a separate description. Operators should use the vendor notice, not this article, as the remediation record.

Why one scheduler bug can become many compromises

An enterprise scheduler usually holds credentials for target hosts, database connections, application accounts, or service integrations. It also knows which jobs run where. An attacker who can invoke agent operations can potentially turn the scheduler into a trusted command broker. Defenses placed in front of the WCC user interface matter less if the agent control channel accepts requests without verifying the caller.

The log-related issues create a second path. A failed login that stores a recoverable password weakens the assumption that failure logs contain only diagnostic metadata. A log with active session identifiers can expose a still-useful admin session even when no password was recorded. These are not separate from the execution flaws. They can provide the credentials needed to reach them.

The hard-coded key is another trust-transfer problem. A key embedded in a product cannot be unique to one customer. If the key protects an agent-management path, changing the WCC's external TLS certificate or placing a web application firewall in front of the login page does not replace the need to fix the key-handling defect.

A response plan for the control plane

Treat the WCC and every managed System Agent as one incident-response scope. First identify WCC versions, exposed web interfaces, and agent registrations. Restrict WCC access to administrative networks or a VPN while the vendor update is prepared. Do not expose the management tier directly to the public internet.

Next collect and protect WCC logs before rotating credentials. Search for login failures that contain password-like values, session identifiers, unusual export requests, or requests to agent-control functions from unexpected sources. Because log files may contain secrets, preserve them under restricted access and avoid copying them into general-purpose ticketing systems.

Rotate scheduler, agent, database, and integration credentials after the fix is deployed. Review job definitions for changes that add command execution, new target hosts, or altered credentials. Confirm that agent identities accept operations only from the intended WCC instances. If the scheduler supports separate service accounts for viewing jobs and changing them, remove administrative rights from accounts that only need read access.

Finally, test the trust path. A patched login page is not enough. Verify that an unauthenticated request cannot reach agent control or configuration endpoints, that export features reject arbitrary destinations, and that failed logins cannot reproduce recoverable secrets. Broadcom's advisory is the authoritative list of affected releases and fixes. The operational lesson is broader: a job scheduler is a privileged automation system, and its API, logs, keys, and agents all belong in the same security model.

References

Top comments (0)