InternLM MindSearch Planner Agent Code Injection Flaw

A vulnerability has been found in InternLM MindSearch 0.1.0. This issue affects the function ExecutionAction.run of the file mindsearch/agent/graph.py of the component Planner Agent. The manipulation of the argument inputs leads to code injection. The attack can be initiated remotely. The exploit has been disclosed to the public and may 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 323matched 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

CVE-2026-105135 is a critical code injection vulnerability (CVSS 10.0) in InternLM MindSearch 0.1.0's Planner Agent component, specifically within the ExecutionAction.run function in mindsearch/agent/graph.py. An attacker can manipulate the inputs argument to inject arbitrary code that executes remotely without authentication. This vulnerability affects any organization deploying MindSearch for agent-based planning and reasoning tasks, particularly those in research, development, or production environments relying on this framework for autonomous decision-making. The lack of vendor response and public disclosure of the exploit significantly increases risk exposure.

Casky's 323 mapped skills detect the attack patterns underpinning this vulnerability by analyzing behaviors aligned with TA0001 (Initial Access) and TA0043 (Reconnaissance). Practitioners would observe detections focused on CWE-74 (Improper Neutralization of Special Elements in Output) and CWE-94 (Improper Control of Generation of Code) through Claude's extended reasoning capabilities. The platform would flag suspicious input manipulation patterns flowing into the ExecutionAction.run function, identify code injection payloads in agent inputs, and correlate these with command execution artifacts. Security teams would see findings highlighting unvalidated input processing in the Planner Agent, untrusted data flowing to code execution contexts, and anomalous agent behavior following malformed input submission—enabling rapid threat hunting and remediation before exploitation occurs.

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 →

323 Casky skills cover this attack pattern

These skills use Claude AI's reasoning model to surface findings in the same attack categories as CVE-2026-105135.

Abuse of Public-Facing API: Mobile API Abuse

mobile security · low

Run

abusing-dpapi-for-credential-access

red teaming · high

Run

abusing-shadow-credentials-for-privesc

red teaming · high

Run

MITRE ATT&CK Techniques

TA0001 TA0043

Investigate the attack patterns behind CVE-2026-105135

Casky has 323 skills that investigate the attack patterns behind CVE-2026-105135. Run one and get CVSS-scored findings in 3 minutes.

Run the skill that detects this →

X Instagram LinkedIn