CVE-2026-82533: Reliance on Untrusted Inputs in a Security Decision in DeepSeek DeepSeek Harness
Description
DeepSeek Harness versions before 0.1.2-alpha.1 contain an authentication bypass vulnerability that allows unauthenticated access to its local HTTP agent-control API by trusting a client-supplied loopback Host header instead of validating the actual TCP connection origin. This flaw enables confined processes to escape OS sandbox restrictions and escalate privileges locally. If the API port is exposed externally via tunnels or proxies, remote attackers can exploit this vulnerability to create sessions, execute arbitrary commands, and exfiltrate stored conversation transcripts without credentials.
CVSS v4.0
Score 9.4critical
Affected software
DeepSeek
DeepSeek Harness
pkg:github/DeepSeek HarnessRun on your own infrastructure? Check whether these packages are installed with threat-finder — our free open-source scanner.
AI-Powered Analysis
Machine-generated threat intelligence
Technical Analysis
CVE-2026-82533 is an authentication bypass vulnerability in DeepSeek Harness prior to version 0.1.2-alpha.1. The vulnerability arises because the software accepts a client-supplied loopback Host header to authenticate access to its local HTTP agent-control API, rather than verifying the true origin of the TCP connection. This allows an attacker with local confined process access to bypass sandbox restrictions and escalate privileges. Additionally, if the API port is exposed externally through mechanisms such as SSH tunnels, forward or reverse proxies, remote attackers can exploit the same flaw to create sessions, execute arbitrary commands, and steal stored conversation data without needing credentials.
Potential Impact
The vulnerability allows unauthenticated attackers to bypass authentication controls on the local HTTP agent-control API. Locally, this can lead to sandbox escape, privilege escalation to unconfined execution, and disabling of approval prompts. Remotely, if the API port is exposed, attackers can fully compromise the system by creating sessions, executing arbitrary commands, and exfiltrating sensitive data such as conversation transcripts. The CVSS 4.0 score of 9.4 reflects the critical severity and the high impact on confidentiality, integrity, and availability.
Mitigation Recommendations
A fix is available in DeepSeek Harness version 0.1.2-alpha.1 and later. Users should upgrade to version 0.1.2-alpha.1 or newer to remediate this vulnerability. Until patched, ensure that the local HTTP agent-control API port is not exposed externally via tunnels, SSH forwarding, or proxies to reduce remote attack risk.
Technical Details
- Data Version
- 5.2
- Assigner Short Name
- VulnCheck
- Date Reserved
- 2026-08-29T17:20:57.082Z
- Cvss Version
- 4.0
- State
- PUBLISHED
Threat ID: 6aa040f6acd9273b49f5b515
Added to database: 09/08/2026, 17:08:06 UTC
Last enriched: 10/07/2026, 21:56:49 UTC
Last updated: 10/24/2026, 03:05:11 UTC
Views: 205
Community Reviews
0 reviewsCrowdsource mitigation strategies, share intel context, and vote on the most helpful responses. Sign in to add your voice and help keep defenders ahead.
Want to contribute mitigation steps or threat intel context? Sign in or create an account to join the community discussion.
Actions
Updates to AI analysis require Pro Console access. Upgrade inside Console → Billing.
Need more coverage?
Upgrade to Pro Console for AI refresh and higher limits.
For incident response and remediation, OffSeq services can help resolve threats faster.
Latest Threats
Check if your credentials are on the dark web
Instant breach scanning across billions of leaked records. Free tier available.