Agent manipulation via context-compaction prompt injection
Published Oct 16, 2026 · Updated Oct 16, 2026
Prompt injection in DeepSeek Harness commit c291e796 allows attackers to influence agent behavior via crafted external content. Context compaction replaces older conversation history with a user-role summary, but the precise trust-boundary failure is not disclosed in the accessible record. Exploitation involves crafted external content reaching context compaction; downstream command execution, data theft, and required privileges are not established.
Summary
What happened
Prompt injection in DeepSeek Harness commit c291e796 allows attackers to influence agent behavior via crafted external content. Context compaction replaces older conversation history with a user-role summary, but the precise trust-boundary failure is not disclosed in the accessible record. Exploitation involves crafted external content reaching context compaction; downstream command execution, data theft, and required privileges are not established.
The record
- CVE
- CVE-2026-95183
- Published
- Oct 16, 2026
- Updated
- Oct 16, 2026
- Vendor
- DeepSeek AI
- Product
- DeepSeek Harness
- Classifications
- Unavailable
- Attack vector
- Unavailable
- Privileges
- Unavailable
Timeline
How it unfolded
- Oct 16, 2026CVE publishedPublication date reported by the CVE source.
- Oct 16, 2026Record updatedLatest update available in the CVE record.
Exploitability
Present is not the same as exploitable
Compare your product and version with the public record. A matching version still requires validation against your environment.
Is a vulnerable build present?
Compare these published version ranges with your installed build and any vendor patches.
- Affected versionversion=n/a
What conditions does exploitation require?
What is affected?
Published CVSS scores
No CVSS assessment is available in this record.
CVSS describes severity. EPSS estimates exploitation probability.
Attacks
What attackers are doing with it
Daily unique IPs observed by Shadowserver honeypots for known exploited vulnerabilities (KEVs). Missing observations do not establish an absence of attacks.
Activity data is delayed. The latest report is Sep 28, 2026.
Weakness, pattern, technique
No sourced classifications are available.
Public exploit references
No public exploit references are available in this record.
Labels summarize the accepted research assessment. They do not indicate a test against your environment.
Technologies
Your stack
See the directory against your own environment.
Your stack
Check the software in your environment
Get started to see how Hinoki identifies affected software and validates exploitability in your environment.
Get started