Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessCompromise or obtain control of enrolled host
-
DeliveryCraft malicious agent-reported value
-
ExploitSubmit value via unauthenticated Okta conditional access endpoint
-
ExecutionSQL injection extracts session tokens from Fleet database
-
PersistReplay admin session token via Fleet API
-
ImpactExecute arbitrary scripts on all enrolled hosts
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V1 - Encoding and Sanitization Parameterized queries, prepared statements
MITRE ATT&CK
2 techniques across 2 tactics: Initial Access, Persistence The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Requires Fleet Premium with the Okta conditional access integration explicitly configured and enabled - instances without this feature are unaffected. The attacker must control at least one host already enrolled in the Fleet instance; this is the minimum foothold required, corresponding to the product's lowest-privilege role. The vulnerable request path itself is unauthenticated (the host agent communicates without user-level authentication), but the precondition of having an enrolled host limits exploitation to parties who have already compromised at least one managed endpoint. Exploitation is not limited to a specific operating system for the attacker's host. What limits scope: the Okta conditional access feature must be active (non-default for most deployments), and initial host enrollment must already be achieved - cold external attackers with no enrolled host cannot trigger the injection. |
| Impact | The CVSS 4.0 vector (AV:N/AC:L/AT:P/PR:L/UI:N/VC:H/VI:H/VA:N) accurately captures the primary risk profile: network-reachable, no user interaction required, but gated behind a specific attack prerequisite (AT:P) - Fleet Premium with Okta conditional access enabled. PR:L reflects that the attacker must control at least one enrolled host, which is the product's lowest-privilege entry point. The confidentiality (VC:H) and integrity (VI:H) impacts are high because the Fleet database stores session tokens for all users including administrators. The vendor's vector scores SC/SI/SA as N, but this understates the lateral impact: successfully replaying a global admin session token enables execution of scripts across all enrolled hosts, representing a substantial subsequent-system impact. Real-world priority is high for any Fleet Premium deployment with Okta conditional access, and low for all others - this is not a spray-and-pray vulnerability. No CISA KEV listing and no public POC have been identified, reducing the urgency slightly for organizations that can verify the feature is disabled. |
| Exploit ScenarioDerived | An attacker with control of a single enrolled host - gained, for example, through malware infection, physical access, or a compromised developer machine - manipulates the values that the host's osquery agent reports to the Fleet server. When Fleet's Okta conditional access endpoint processes these values in an unsanitized database query, the attacker extracts session tokens for all Fleet users including global administrators. The attacker then replays a global admin token via the Fleet API and uses the administrative script-execution feature to run arbitrary commands on every other host enrolled in the Fleet instance. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv2 – Aug 26, 2026
SQL injection in Fleet's Okta conditional access integration allows an attacker controlling any single enrolled host to read or modify arbitrary Fleet database contents, including stored session tokens. Disclosed tokens can be replayed to impersonate a global administrator, enabling remote code execution on all managed hosts. The CVSS 4.0 score of 7.6 (AT:P) reflects that exploitation requires Fleet Premium with the Okta conditional access feature enabled, but within that deployment profile the impact chain extends to full fleet compromise. No public exploit identified at time of analysis; vendor-released patch is available as fleet-v4.86.2.
Earlier versions of this analysis (1)
SQL injection in Fleet Premium's Okta conditional access integration allows an attacker controlling a single enrolled host to exfiltrate arbitrary Fleet database contents - including stored session tokens - and replay those tokens to gain global administrator access, ultimately achieving remote code execution on all enrolled hosts. The vulnerable endpoint is an unauthenticated request path that incorporates a host-supplied value into a database query without parameterization, meaning any party who controls even the lowest-privileged entity in the system (an enrolled host agent) can influence the query. No public exploit has been identified at time of analysis, but a patch is available in fleet-v4.86.2.
RemediationAI
Vendor-released patch: fleet-v4.86.2. Upgrade to fleet-v4.86.2 or later, available at https://github.com/fleetdm/fleet/releases/tag/fleet-v4.86.2, which resolves the SQL injection by parameterizing the host-supplied value in the affected query. If an immediate upgrade is not feasible, the vendor-recommended workaround is to disable the Okta conditional access integration entirely until the patched version is deployed - this removes the vulnerable code path with no impact on core Fleet functionality, though Okta-based conditional access enforcement will cease. No other compensating controls are specified by the vendor. After patching, administrators should rotate all Fleet session tokens and review database audit logs for anomalous queries or unexpected token use, as prior exploitation may have left persistent access through replayed credentials.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 3 packages in go, euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
Fleet is an open-source, Go-based MDM and device management platform (pkg:go/github.com/fleetdm/fleet) that integrates with osquery agents running on enrolled hosts. The vulnerable component is an unauthenticated HTTP endpoint servicing the Okta conditional access integration. This endpoint consumes a value supplied by the host's own osquery agent - a value that is stored verbatim and then interpolated directly into a database query without parameterization. The root cause is CWE-89 (Improper Neutralization of Special Elements in SQL Commands), where host-agent-supplied input reaches a SQL query constructor without binding. Because any party controlling a host can manipulate what the agent reports, the trust boundary between enrolled host and the Fleet server's database is effectively eliminated when this feature is active.
Know when the next one hits your stack
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:N/AC:L/AT:P/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N
AC:H reflects the Fleet Premium plus Okta conditional access prerequisite; PR:L reflects the enrolled-host requirement; S:C captures privilege escalation to global admin enabling cross-host RCE.
Primary rating from Vendor (https://github.com/fleetdm/fleet).
CVSS VectorVendor: https://github.com/fleetdm/fleet
Lifecycle Timeline
2DescriptionCVE.org
Summary
A SQL injection vulnerability in Fleet's Okta conditional access integration could allow an attacker who controls a single enrolled host to read or modify arbitrary data in the Fleet database, including stored session tokens. Disclosed session tokens may be replayed to act as a global administrator, which on a managed fleet leads to remote code execution on enrolled hosts.
Impact
When Fleet Premium with Okta conditional access is configured, an unauthenticated request path that supports the conditional access integration uses a host-supplied value in a database query without proper parameterization. Because the value is reported by the host's own agent and stored verbatim, any party that controls one enrolled host (the lowest-privilege role in the product) can influence the query.
Successful exploitation could allow:
- Disclosure of arbitrary database contents, including credentials and session tokens.
- Replay of disclosed session tokens to gain global administrator access.
- Subsequent actions available to a global administrator, including running scripts on enrolled hosts.
This issue requires Fleet Premium with the Okta conditional access integration enabled. It does not affect instances where Okta conditional access is not configured.
Workarounds
If an immediate upgrade is not possible, disable the Okta conditional access integration until the patched version is deployed.
For more information
If you have any questions or comments about this advisory:
Email us at [security@fleetdm.com](mailto:security@fleetdm.com)
Join #fleet in osquery Slack
Credits
We thank @fuzzztf for responsibly disclosing this issue.
Executive ViewAI
A SQL injection vulnerability in Fleet's Okta conditional access integration creates a pathway for attackers with access to any single managed endpoint to assume administrative control over the entire endpoint management infrastructure and execute code on all managed devices. Organizations using Fleet Premium with this feature enabled face potential exposure of all managed systems and stored credentials if unpatched. This poses an immediate operational and compliance risk to the security posture of any organization deploying this configuration.
Apply vendor patch.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
SQL injection exploits the way applications construct database queries by mixing user input directly into SQL statements.
Remote Code Execution represents the critical moment when an attacker successfully runs arbitrary code on a target system without physical access.
EUVD-2026-66673
GHSA-7q96-f8xw-jv5j
NVD
CVE.org
EPSS
Exploit-DB