CVE-2026-96620: Claude Agent SDK

The National Vulnerability Database (NVD) records that before 0.2.121 the Claude Agent software development kit (SDK) for Python passed the resume option to the Claude CLI without validation, so an application passing untrusted input as the session ID could let an attacker inject an --mcp-config flag and run arbitrary commands at CLI startup outside the SDK's permission controls; fixed in 0.2.121.

The record

Kind

Vulnerability

Source record

CVE-2026-96620: Claude Agent SDK is a Python SDK for Claude Code. Prior to 0.2.121, the resume option is passed to the Claude (National Vulnerability Database record)

Record published

2026-10-16

Posted here

2026-10-19

World

Software agents · Enterprise work agents

Controls concerned

Machine-enforceable scope C-M2; Containment and isolation C-E5

Checked

Checked against its record on 2026-10-22 by a reader who did not draft it; report verification/reports/SIG-2026-10-19-003-2026-10-22.md.

A signal's controls are the ones its record points at, the exposure; a case's controls are the ones its records show were absent or failed, so the two can differ.

Corrections

Changes to what this signal says, with the reason. Every earlier version is kept.

• 2026-10-22: Summary. Plain-language rule: abbreviations spelled out at first use (NVD, MCP, SDK). No claim changed. 22 October 2026.
  Show the change

  Before: NVD records that before 0.2.121 the Claude Agent SDK for Python passed the resume option to the Claude CLI without validation, so an application passing untrusted input as the session ID could let an attacker inject an --mcp-config flag and run arbitrary commands at CLI startup outside the SDK's permission controls; fixed in 0.2.121.

  After: The National Vulnerability Database (NVD) records that before 0.2.121 the Claude Agent software development kit (SDK) for Python passed the resume option to the Claude CLI without validation, so an application passing untrusted input as the session ID could let an attacker inject an --mcp-config flag and run arbitrary commands at CLI startup outside the SDK's permission controls; fixed in 0.2.121.

What a signal is. A signal records what a primary record says; it assigns no fault. It is lighter than a case: one dated item from one record, tagged with the world, sector and controls it touches. It becomes a case when it meets the bar for one.

Newer: CVE-2026-77474: destructive_command_guard (dcg)
Older: CVE-2026-88755: Z.ai ZCode connects to repository-declared MCP servers