CVE-2026-96620Public exploitNetwork attack vectorHinoki check available

Command execution via resume argument injection

Published Oct 16, 2026 · Updated Oct 16, 2026

Argument injection in Anthropic Claude Agent SDK for Python before 0.2.121 allows attackers to execute commands via resume IDs. SubprocessCLITransport._build_command() passes the resume value as a separate argument, letting dash-leading input become a CLI flag. Exploitation requires application-supplied untrusted session input; injected MCP configuration starts commands outside SDK permission controls.

CVSS severity9.2
Critical
EPSS probability0.58%
Next 30 days · Oct 19, 2026
Known exploitationUnconfirmed
Based on sourced intelligence
Hinoki checkAvailable
Coverage for this vulnerability

See if you're affected

Explore vulnerability checks for your environment with Hinoki.

Get started

Summary

What happened

Argument injection in Anthropic Claude Agent SDK for Python before 0.2.121 allows attackers to execute commands via resume IDs. SubprocessCLITransport._build_command() passes the resume value as a separate argument, letting dash-leading input become a CLI flag. Exploitation requires application-supplied untrusted session input; injected MCP configuration starts commands outside SDK permission controls.

The record

CVE
CVE-2026-96620
Published
Oct 16, 2026
Updated
Oct 16, 2026
Vendor
Anthropic
Product
Claude Agent SDK for Python
Classifications
CWE-20, CWE-88, CWE-78, T1203
Attack vector
network
Privileges
unauthenticated

Timeline

How it unfolded

  1. Oct 16, 2026CVE publishedPublication date reported by the CVE source.
  2. 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.

  1. Affected versionversion=< 0.2.121

What conditions does exploitation require?

Attack vectornetwork
Required privilegesunauthenticated

What is affected?

Anthropic · Claude Agent SDK for Pythonversion=< 0.2.121

Published CVSS scores

9.2AnthropicCVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N

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.

Daily unique IPsNo honeypot observations are available for this CVE in the selected window.

No observations available

Sep 22, 2026Sep 28, 2026
Latest reporting daySep 28, 2026
Latest daily unique IPsUnavailable
Prior 30-day averageUnavailable
SourceShadowserver honeypots (KEV)
Vectornetwork
Privilegesunauthenticated
Known exploitationUnconfirmed
Public exploitPublished

Weakness, pattern, technique

CWE-20Improper Input Validation
CWE-88Improper Neutralization of Argument Delimiters in a Command ('Argument Injection')
CWE-78Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')
T1203Exploitation for Client Execution

Public exploit references

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