Splunk Enterprise Auth Bypass (CVE-2026-76536) and SOAR RCE: Patch Now

Rescana advisory card: Splunk Enterprise authentication bypass CVE-2026-76536 and Splunk SOAR code execution flaw CVE-2026-131253. Patch now.

Status: On November 18, 2026, Splunk published its November security advisories. Three of them include flaws that Splunk scores above 9, and this post covers two; the third, for Splunk Enterprise Security, is summarised below. SVD-2026-1103 fixes 38 vulnerabilities in self-managed Splunk Enterprise, including CVE-2026-76536 (CVSS 3.1 9.4) and CVE-2026-76535 (CVSS 3.1 9.1), both of which Splunk says an unauthenticated user can reach because Splunk Enterprise accepts an empty shared secret. SVD-2026-1108 fixes 59 vulnerabilities in Splunk SOAR, including CVE-2026-131253 (CVSS 3.1 9.9), a code-execution flaw in the App Wizard that needs an authenticated user with specific permissions. Splunk rates all three Critical. The fixes are Splunk Enterprise 10.6.1, 10.4.5, 10.2.9, 10.0.12 and 9.4.17, and Splunk SOAR 8.8.0, all of which were already downloadable before the advisories went up. For the two Splunk Enterprise flaws, Splunk also gives a pass4SymmKey workaround in server.conf. CVE-2026-76536 is not in CISA's Known Exploited Vulnerabilities (KEV) catalog. CVE-2026-76535 is not in CISA's Known Exploited Vulnerabilities (KEV) catalog. CVE-2026-131253 is not in CISA's Known Exploited Vulnerabilities (KEV) catalog. We checked catalogVersion 2026.11.18 at 02:02 EST on November 20, 2026. Splunk does not report exploitation, and we found no public proof-of-concept code. We last checked all sources between 02:02 and 02:04 EST on November 20, 2026.

Executive Summary

Splunk Enterprise nodes call each other's REST endpoints on the splunkd management port and authenticate those calls with a shared secret called pass4SymmKey (Rescana's summary of Splunk's documentation). Splunk says affected releases accept an empty secret. For CVE-2026-76535 that lets an unauthenticated user run searches through RapidDiag, Splunk's diagnostic collection app, "as an authenticated user." For CVE-2026-76536 it lets an unauthenticated user read and change clustering configuration on a standalone instance that does not use clustering at all.

What matters for planning:

  • No login is needed for the two Splunk Enterprise flaws. Both CVSS vectors read network-reachable, low complexity, no privileges, no user interaction. CISA's enrichment record for CVE-2026-76535 rates it automatable.
  • The SOAR flaw needs an account, but not an administrator. CVE-2026-131253 needs a role with the Apps edit, Containers edit and Playbooks execute permissions, and the result is code execution as the account that runs Splunk SOAR.
  • The fixes were public weeks before the advisories. Splunk Enterprise 10.6.1 shipped on October 16, 2026, the other four fixed releases on October 30, and SOAR 8.8.0 on November 4, so each build has been available for at least two weeks (Rescana's reading).
  • There is no report of exploitation. Splunk's advisories do not mention it, and CISA's enrichment records say "none."

Who should act first (Rescana guidance). Teams running self-managed Splunk Enterprise whose management port is reachable from networks they do not fully control, instances that run with clustering turned off, such as heavy forwarders, deployment servers and single-server deployments, and any Splunk SOAR (On-premises) deployment where many users hold the Apps edit permission. Then the managed security and SOC providers that run Splunk on your behalf.

Technical Information

CVE-2026-76535, as Splunk describes it. "In Splunk Enterprise versions below 10.6.1, 10.4.5, 10.2.9, 10.0.12, and 9.4.17, an unauthenticated user could run arbitrary Search Processing Language (SPL) searches through RapidDiag task collection as an authenticated user. With this access, the user could read or modify data and configuration available to that authenticated user. The vulnerability is possible because Splunk Enterprise accepts an empty shared secret when it authenticates RapidDiag task-collection requests." Splunk classifies it as CWE-287 (improper authentication) and scores it CVSS 3.1 9.1, CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:N.

CVE-2026-76536, as Splunk describes it. "In Splunk Enterprise versions below 10.4.5, 10.2.9, 10.0.12, and 9.4.17, an unauthenticated user could use an empty clustering shared secret to read and modify clustering configuration through the cluster configuration Representational State Transfer (REST) API endpoint on a standalone instance. This could expose clustering configuration, persist attacker-controlled clustering settings, and disrupt clustering operations. The vulnerability is possible because Splunk Enterprise accepts the empty clustering shared secret as a shared-authentication candidate on a standalone instance when clustering is turned off." Also CWE-287, scored CVSS 3.1 9.4, CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:H/A:H.

Two details that are easy to miss (Rescana's reading). First, CVE-2026-76536 names the 10.4, 10.2, 10.0 and 9.4 lines but not 10.6; its CVE record's version list matches. CVE-2026-76535 includes 10.6. Second, Splunk describes CVE-2026-76536 as affecting a "standalone instance" with clustering turned off. In a distributed deployment that can include more than single-server installs: deployment servers, license managers, the monitoring console, heavy forwarders and search heads that do not search an indexer cluster can all run with indexer clustering off. Indexer cluster managers and peers are not the case Splunk describes.

Two related flaws in the same advisory. Splunk lists two more shared-secret bugs at CVSS 3.1 7.5:

  • CVE-2026-131192 (High): on a standalone instance or a search head that is not an indexer cluster member, "an unauthenticated user could access live event data from the indexing pipeline" because Splunk Enterprise "accepts an empty shared key during service authentication." It affects the same five lines as CVE-2026-76535, 10.6 included.
  • CVE-2026-131190 (High): "certain RapidDiag REST API endpoints accept a factory-default shared key for node authentication," exposing RapidDiag process metadata and log file inventory to an unauthenticated user. Splunk credits Jason Storm and Gabriel Nitu of Splunk.

Splunk names no external finder for CVE-2026-76535 or CVE-2026-76536.

CVE-2026-131253 (Splunk SOAR), as Splunk describes it. "In Splunk SOAR versions below 8.8.0, an authenticated user that holds a role with the Apps edit, Containers edit, and Playbooks execute permissions could use the App Wizard to execute arbitrary code as the user account that runs Splunk SOAR." The cause: "the App Wizard does not validate user-supplied identifiers before incorporating them into generated connector code." CWE-94, CVSS 3.1 9.9, CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H.

The second SOAR flaw above 9. CVE-2026-131287 (CVSS 3.1 9.6, CWE-863): a user with the Playbooks view, edit, execute and delete permissions "could cause existing cross-repository playbook references to run attacker-controlled playbook code," because SOAR "can bind existing cross-repository playbook references to a different source control repository when a deleted repository name is reused." Several other SOAR flaws in SVD-2026-1108, scored 7.2 to 8.8, also end in code execution for users who can edit apps or system settings.

Scores and who set them. Cisco, Splunk's parent company, published the CVE records as the CVE Numbering Authority (CNA) on November 19, 2026, between 14:55 and 14:56 EST. NVD lists all four records above 9 as "Received" with the CNA's scores and has published no assessment of its own. Rescana's Critical severity field for this post adopts the CNA scores and nothing else.

Exploitation and prioritization signals, as published.

Signal Value Source and time
Vendor statement Splunk's advisories do not mention exploitation SVD-2026-1103 and SVD-2026-1108, checked 02:02 EST on November 20, 2026
Catalog status, CVE-2026-76536 CVE-2026-76536 is not in CISA's Known Exploited Vulnerabilities (KEV) catalog. catalogVersion 2026.11.18, 1,786 entries, checked 02:02 EST on November 20, 2026
Catalog status, CVE-2026-76535 CVE-2026-76535 is not in CISA's Known Exploited Vulnerabilities (KEV) catalog. Same check
Catalog status, CVE-2026-131253 CVE-2026-131253 is not in CISA's Known Exploited Vulnerabilities (KEV) catalog. Same check
CISA enrichment (SSVC), CVE-2026-76535 Exploitation: none; Automatable: yes; Technical Impact: total CISA ADP entry dated November 19, 2026
CISA enrichment (SSVC), CVE-2026-131253 Exploitation: none; Automatable: no; Technical Impact: total CISA ADP entry dated November 19, 2026
CISA enrichment (SSVC), CVE-2026-131287 Exploitation: none; Automatable: no; Technical Impact: total CISA ADP entry dated November 19, 2026
CISA enrichment (SSVC), CVE-2026-76536 No CISA entry on the record yet CVE.org, checked 02:02 EST on November 20, 2026
EPSS No score yet for any of the three FIRST EPSS API, checked 02:02 EST on November 20, 2026
Public proof of concept None reported by any source we opened Splunk, CVE.org, NVD, CERT Santé, SecurityWeek; checked 02:02 to 02:03 EST on November 20, 2026

What we could not establish. Splunk does not say whether an empty pass4SymmKey is the out-of-the-box state for the services involved, which RapidDiag endpoints are affected, or whether the RapidDiag flaw needs anything beyond network access to the management port. Splunk's FAQ says it "does not provide additional information about the specifics of vulnerabilities beyond what it discloses." CERT Santé, the French health-sector CERT, rated CVE-2026-76535 on November 19 as not actively exploited with no proof of concept in open sources (Rescana's conversion from French), and SecurityWeek reports that Splunk "makes no mention of any of these security defects being exploited in the wild."

Affected Product Versions

Product and line Affected, per Splunk Fixed release Fixed release published
Splunk Enterprise 10.6 10.6.0 builds (the table shows "10.6.0.5"); 7 of the 38 CVEs, including CVE-2026-76535 and CVE-2026-131192 but not CVE-2026-76536 10.6.1 October 16, 2026
Splunk Enterprise 10.4 10.4.0 to 10.4.4 10.4.5 October 30, 2026
Splunk Enterprise 10.2 10.2.0 to 10.2.8 10.2.9 October 30, 2026
Splunk Enterprise 10.0 10.0.0 to 10.0.11 10.0.12 October 30, 2026
Splunk Enterprise 9.4 9.4.0 to 9.4.16 9.4.17 October 30, 2026
Splunk SOAR Below 8.8.0 8.8.0 November 4, 2026

The release dates come from Splunk's release notes. Splunk Enterprise 10.6.2 followed on November 19, 2026.

How to read the 10.6 row (Rescana's reading). Starting with 10.6, Splunk moved to a four-segment version format, and its 10.6 release notes give "10.6.0.5" as the example. The CVE records for the seven CVEs that name 10.6, including CVE-2026-76535, list the 10.6 line as affected below 10.6.1, and Splunk's Universal Forwarder advisory for the same releases writes the range as "10.6.0 to 10.6.0.5," so treat every 10.6 build below 10.6.1 as affected.

Splunk Cloud Platform versus self-managed Splunk Enterprise. The Product Status table of SVD-2026-1103 lists only Splunk Enterprise 9.4 through 10.6, not Splunk Cloud Platform. Splunk's advisory FAQ says that table "explicitly lists which products and versions are affected," and that Splunk Cloud Platform customers "will receive a notification from Splunk about scheduling your update." Splunk makes no other statement about Splunk Cloud Platform in these advisories. Rescana's reading: these action items are for self-managed Splunk Enterprise. If you run Splunk Cloud Platform, ask Splunk to confirm in writing whether your stack was affected, and check any self-managed heavy forwarders or deployment servers beside it, which are Splunk Enterprise installs.

Splunk SOAR (On-premises) versus Splunk SOAR (Cloud). SVD-2026-1108 lists "Splunk SOAR" below 8.8.0 without splitting the two. Splunk's SOAR (Cloud) release notes say 8.8.0 "was released for General Availability on November 4, 2026." Rescana's reading: on-premises customers must upgrade themselves; SOAR (Cloud) customers should confirm with Splunk that their tenant runs 8.8.0 or later.

Universal Forwarder. SVD-2026-1103 does not list the Universal Forwarder. A separate advisory, SVD-2026-1105, updates third-party packages in Universal Forwarder releases with the same version numbers, rated Medium by Splunk.

Older releases. Splunk's FAQ says it "has not tested or verified the impact of vulnerabilities on versions it does not currently support." Treat any Splunk Enterprise release below 9.4 as unpatched and plan a move to a fixed line (Rescana guidance). Separately, Splunk's October Node.js advisory (SVD-2026-1015, updated November 17) says Splunk Enterprise 10.0 bundles an affected Node.js runtime and recommends 10.2 or higher; weigh that before choosing 10.0.12.

Operating systems. SVD-2026-1103 names no operating system for these four flaws, and Splunk's FAQ says an advisory that lists none "applies to all operating systems upon which Splunk supports running the affected software." Splunk's RapidDiag documentation says the app ships with Splunk Enterprise 8.1.1 and later and runs on Linux only; Splunk does not use that to narrow the advisory, so neither do we (Rescana's reading).

Other November 2026 Splunk Advisories

Splunk published twelve advisories for November between November 11 and 19. Our reading of the rest (Rescana's reading):

  • SVD-2026-1106, Splunk Enterprise Security and Splunk App for PCI Compliance, fixed in 8.7.0 (42 CVEs). Mentioned because Enterprise Security runs on Splunk Enterprise search heads and belongs in the same change window. Its top flaw, CVE-2026-131202 (CVSS 3.1 9.1), lets a user with the "ess_admin" role "execute arbitrary code on the search head." Splunk's workaround is to turn off or remove the app.
  • SVD-2026-1102, Java (CVE-2026-114938). Splunk lists Splunk Enterprise, Splunk Cloud Platform, the Universal Forwarder, Splunk SOAR and SOAR Cloud as unaffected, and says Splunk IT Service Intelligence and Splunk DB Connect "use a separately installed Java runtime," which self-managed customers should update.
  • The rest, in brief. SVD-2026-1104, 1105, 1107 and 1109 are third-party package updates that ship in the same fixed releases; the Universal Forwarder and Enterprise Security are upgraded separately from Splunk Enterprise. SVD-2026-1110 (Splunk App for Stream, 8.1.7) and SVD-2026-1111 (Splunk AI Assistant, 2.0.0) are app updates scored up to 7.5 and 8.1, each with a turn-off-or-remove workaround; update them if you run them. SVD-2026-1112 (NGINX) lists no affected product yet; Splunk "is investigating its product line."
  • SVD-2026-1101. The advance notice of November 11, used for the timeline below. It also listed Splunk SOAR Connectors; as of 02:02 EST on November 20, 2026, Splunk's advisory list and RSS feed show no November advisory for SOAR Connectors.

Timeline (EST)

  • October 16 to November 4, 2026: the fixed releases ship, dated in the Affected Product Versions table.
  • November 11: Splunk's advance notice (SVD-2026-1101) says it will publish advisories on November 18 for vulnerabilities fixed in these releases.
  • November 18: Splunk publishes SVD-2026-1103, SVD-2026-1108 and the other November advisories.
  • November 19, 05:41: SecurityWeek reports the advisories.
  • November 19, 14:55 to 14:56: the CVE records are published, with NVD entries following from 15:17. Splunk Enterprise 10.6.2 is released the same day.
  • November 19, 23:56: CISA's enrichment is added to the CVE-2026-76535, CVE-2026-131253 and CVE-2026-131287 records, with Exploitation "none."

How to Find Your Exposure

These steps are Rescana guidance, not Splunk's.

  1. Inventory every Splunk Enterprise instance and its version, not just the search heads: indexers, cluster managers, deployment servers, license managers, the monitoring console and heavy forwarders. On each host, $SPLUNK_HOME/bin/splunk version prints the release; anything below the fixed release for its line is in scope.
  2. Find out who can reach the management port. Splunk's REST API reference addresses endpoints as https://<host>:<mPort>/services/... and uses 8089 in its examples; check the port your instances use. From a host outside your Splunk network, confirm whether it answers at all. It should be reachable only from other Splunk nodes and your administrators.
  3. Read the effective pass4SymmKey settings with $SPLUNK_HOME/bin/splunk btool server list clustering --debug and $SPLUNK_HOME/bin/splunk btool server list general --debug, and look for a pass4SymmKey line with nothing after the equals sign, or one whose --debug source is etc/system/default/server.conf, which means you never set your own key. Splunk's server.conf reference gives [general] a factory default of changeme, and says [clustering] has no default and falls back to [general]. Splunk encrypts a key in server.conf after a restart, so a key you set will not show in plain text.
  4. Check whether the RapidDiag app is installed and enabled. Splunk's documentation places its interface under Settings, System, RapidDiag.
  5. Search the splunkd access log (index=_internal sourcetype=splunkd_access, or $SPLUNK_HOME/var/log/splunk/splunkd_access.log) for requests to /services/cluster/config and to RapidDiag endpoints from addresses that are not your Splunk nodes. Splunk does not say when anyone first looked at these flaws, so search as far back as your logs go. October 16, 2026, when the first fixed release shipped, is a sensible minimum.
  6. On standalone instances, check [clustering] in etc/system/local/server.conf for settings you did not make, and its modification time. Splunk says CVE-2026-76536 can "persist attacker-controlled clustering settings."
  7. Review the audit index (index=_audit) for searches you cannot match to a person or a scheduled job, since CVE-2026-76535 runs searches as an authenticated user.
  8. For SOAR, list every role that holds Apps edit, Containers edit and Playbooks execute, and every user in those roles. Then review apps created or changed through the App Wizard, app installs, and source control repositories that were deleted and added again under the same name.
  9. Keep the logs and configuration files you find before you upgrade or restart.

Workaround and Mitigation

Splunk's fix: upgrade. For Splunk Enterprise, Splunk says to upgrade to "versions 10.6.1, 10.4.5, 10.2.9, 10.0.12, and 9.4.17 or higher." For SOAR, "Upgrade Splunk SOAR to version 8.8.0 or higher."

Splunk's workaround for CVE-2026-76535: "Turn off requests authenticated with an empty shared secret by setting pass4SymmKey = <non-empty value> in the server.conf configuration file if you do not use an empty shared secret. Use the same value on each node that participates in the service and restart Splunk Enterprise after the change."

Splunk's workaround for CVE-2026-76536: "If you cannot immediately upgrade to a fixed version, set a strong, unique pass4SymmKey value in the [clustering] stanza in the server.conf configuration file, including on standalone instances. The pass4SymmKey value must contain at least 12 characters. The file is at etc/system/local/server.conf under the Splunk Enterprise installation directory. Restart Splunk Enterprise for the change to take effect."

Splunk's workarounds for the two related flaws: for CVE-2026-131192, "Set pass4SymmKey to a strong value in the server.conf configuration file under the [clustering] stanza"; for CVE-2026-131190, "Turn off or remove the RapidDiag app."

Splunk on workarounds in general: its FAQ calls mitigations "a stopgap measure," and asks customers to "refrain from using them long-term."

Splunk's position for SOAR: no mitigation or workaround for CVE-2026-131253 or CVE-2026-131287. Upgrade.

Where the keys live (Rescana's reading of Splunk's documentation). Splunk's pass4SymmKey page sets keys per server.conf stanza: [clustering] for indexer clusters, [shclustering] for search head clusters, [deployment] for deployment servers and clients, and [general] for "all other services, including licensing." Splunk's CVE-2026-76535 workaround does not name a stanza. Because the documentation assigns services without their own stanza to [general], that is the likely place, but Splunk does not say so (Rescana's reading). Ask Splunk Support which one applies to RapidDiag in your release before you rely on the workaround alone.

Additional steps (Rescana guidance, not Splunk's).

  1. Upgrade rather than rely on keys. The fixed releases have been out since October, and the advisories also close 34 other Splunk Enterprise flaws that a key change does not touch.
  2. Follow Splunk's upgrade order. For an indexer cluster, Splunk requires the manager node first, then the search head tier, then the peer nodes; search head clusters and deployment servers have their own documented steps. See Upgrade an indexer cluster.
  3. Plan key changes across every node that shares the service. Splunk's documentation says instances "must share the same pass4SymmKey to authenticate with one another." Changing the [general] or [clustering] key on one member of a license pool or cluster without the others breaks that connection. CVE-2026-76536 concerns instances with clustering turned off, where the [clustering] key authenticates to no other node, so a value unique to that instance breaks nothing. Splunk's indexer-cluster upgrade guide says clusters have required a non-default key, "the same across all nodes in the cluster," since 6.6; keep that shared key on clustered nodes (Rescana's reading; confirm with Splunk Support).
  4. Record each new key in plain text somewhere safe before you restart. Splunk warns that "it is very difficult to recover the key if you forget it" once it has been encrypted.
  5. Restrict the management port to Splunk nodes and administrator hosts until all your instances run a fixed release, at the network layer and, where it fits, with the acceptFrom setting in the [httpServer] stanza of server.conf. This is a stopgap and does not replace Splunk's fix.
  6. If you do not use RapidDiag, turn it off after the upgrade too, since Splunk's own workaround for CVE-2026-131190 is to remove it.
  7. For SOAR before 8.8.0, cut the number of users who hold Apps edit, Containers edit, Playbooks execute and System Settings edit to the people who need them, and review recently added apps. This narrows who could use CVE-2026-131253 but does not fix it.
  8. Re-run the btool check from exposure step 3 after every upgrade and redeploy, so a configuration push does not bring back an empty key.
  9. If you find requests you cannot explain, treat the instance as potentially compromised. Preserve logs, rotate credentials stored in Splunk (for example in saved searches, alert actions, apps and SOAR assets), and rebuild from a known-good state.

Indicators of Compromise

Splunk, CISA and NVD publish no indicators of compromise, network signatures, source addresses or malware names for these flaws, and none of them reports exploitation or names a threat actor. We list none and invent none.

Signals worth checking in the meantime (Rescana guidance): requests to the management port from addresses outside your Splunk nodes and administrators; requests to /services/cluster/config on instances that do not run clustering, from any address; new or changed [clustering] settings on standalone instances; audit-index searches that no person or scheduled job explains; and, on SOAR, apps or connectors created through the App Wizard that nobody can account for, or unexpected child processes of the SOAR service account. These are generic signs of misuse, not specific to these flaws. Splunk's FAQ points customers to Splunk Enterprise Security Content Updates and research.splunk.com for detections "if available"; we found none for these CVEs as of 02:02 EST on November 20, 2026. We will update this advisory if Splunk or CISA publish indicators.

Third-Party Risk (TPRM) Angle

Splunk often sits with your suppliers: a managed security service provider or MDR vendor running Splunk Enterprise and SOAR for many customers, an outsourced SOC that forwards your logs into its own indexers, or a software supplier that keeps its own logs in Splunk. A Splunk Enterprise instance holds log data, saved credentials and searches across everything it ingests, and a SOAR instance holds credentials and playbooks that act on the systems it is connected to. For a supplier whose management port is reachable, the Splunk Enterprise flaws need no login, so the supplier's user accounts, passwords and multi-factor authentication do not help; only network restrictions and the upgrade do (Rescana's reading).

The suppliers to ask first (Rescana guidance):

  • Managed security, MDR and SOC providers that run Splunk Enterprise or SOAR with your logs or your credentials.
  • Suppliers that receive your log data through forwarders into their own Splunk environment.
  • Any supplier that runs Splunk SOAR with automation connected to your systems.

Send these five questions (Rescana guidance):

  1. Do you run self-managed Splunk Enterprise or Splunk SOAR (On-premises) that stores our data or can act on our systems, and which release is each instance on?
  2. For Splunk Enterprise below 10.6.1, 10.4.5, 10.2.9, 10.0.12 or 9.4.17, or SOAR below 8.8.0, on what date did you upgrade, or apply Splunk's pass4SymmKey workarounds and restart?
  3. Is the splunkd management port of those instances reachable from the internet or from networks shared with other customers?
  4. Have you reviewed splunkd access logs and the audit index for requests to the cluster configuration and RapidDiag endpoints, back to what date, and what did you find?
  5. For Splunk Cloud Platform or SOAR (Cloud), has Splunk confirmed to you in writing whether your stack was affected and when it was updated?

Ask for evidence, not reassurance: the version each instance reports after the upgrade, the date of the change, the log period reviewed and the result. Splunk does not report exploitation, so a normal follow-up date is reasonable, but the two Splunk Enterprise flaws need no login, so do not let answers drift past your next review cycle.

We will update this advisory if Splunk revises SVD-2026-1103 or SVD-2026-1108, reports exploitation, or if CISA adds any of these CVEs to its catalog.

Need help working out which suppliers run self-managed Splunk Enterprise or Splunk SOAR with access to your data? Our team can help you prioritize them and track their answers. Talk to an expert

Sources

Talk to an expert

Talk to Rescana about this advisory — or about how we help TPRM teams keep reviews moving.