Injection Vulnerability in Hermes Agent Context Compression

A vulnerability was identified in NousResearch hermes-agent up to 0.12.0. Affected by this vulnerability is the function _compress_context of the file run_agent.py. The manipulation leads to injection. It is possible to initiate the attack remotely. The exploit is publicly available and might be used. The vendor was contacted early about this disclosure but did not respond in any way.

Casky was already ahead

This CVE exploits attack patterns that Casky's 301matched skills already investigate — long before this vulnerability was disclosed. Claude's reasoning model maps these techniques to MITRE ATT&CK, so practitioners who ran these skills have already seen the threat behaviour in their findings.

Analysis

A critical injection vulnerability exists in NousResearch's hermes-agent library (versions up to 0.12.0) within the _compress_context function of run_agent.py. This flaw allows remote attackers to inject malicious code through context manipulation, potentially leading to unauthorized command execution or data exfiltration. The vulnerability is particularly concerning because hermes-agent is used in AI-driven automation workflows, and the publicly available exploit combined with vendor non-responsiveness creates an immediate risk window for organizations deploying this component in production environments.

Casky's 301 matched security skills enable practitioners to detect attack patterns associated with TA0001 (Initial Access) and TA0043 (Reconnaissance) by analyzing how injected payloads traverse the context compression pipeline. Using Claude's extended reasoning capabilities, practitioners can identify suspicious input patterns in agent logs—such as unusual serialization attempts, encoding anomalies in context data, or unexpected function calls that bypass normal validation. The platform maps detection signals to CWE-74 (Improper Neutralization of Special Elements) and CWE-707 (Improper Neutralization), allowing security teams to construct custom detection rules that correlate context manipulation attempts with downstream execution patterns and flag reconnaissance activity that precedes injection attempts.

Live Threat IntelligenceComing soon

Casky Risk Score

—

EPSS Probability

—

Shodan Exposure

—

GreyNoise Signal

—

Composite risk scoring from EPSS, CISA KEV, Shodan, and GreyNoise — 21 security APIs correlated into a single Casky Risk Score. Coming in Casky Pro. Join early access →

301 Casky skills cover this attack pattern

These skills use Claude AI's reasoning model to surface findings in the same attack categories as CVE-2026-10221.

analyzing-apt-group-with-mitre-navigator

threat intelligence · low

Run

analyzing-campaign-attribution-evidence

threat intelligence · low

Run

analyzing-certificate-transparency-for-phishing

threat intelligence · low

Run

MITRE ATT&CK Techniques

TA0001 TA0043

Investigate the attack patterns behind CVE-2026-10221

Casky has 301 skills that investigate the attack patterns behind CVE-2026-10221. Run one and get CVSS-scored findings in 3 minutes.

Run the skill that detects this →

X Instagram LinkedIn