CVE-2026-82537

Deferred Deferred - Pending Action

Auto-approve Bypass in Roo-Code via Bash Word-Boundary Mismatch

Publication date: 2026-09-08

Last updated on: 2026-09-11

Assigner: VulnCheck

Description

Roo-Code through 3.54.0 contains an auto-approve bypass vulnerability that allows attackers to execute denied shell commands by exploiting a word-boundary mismatch in comment handling between the approval gate's shell parser and bash. Attackers can craft a command string with an allowlisted word immediately followed by a hash character, separator, and denied command to pass the approval gate while bash executes the denied command with the agent's auto-execute privileges on the developer's machine.

CVSS Scores

EPSS Scores

Probability:
Percentile:

Meta Information

Published

2026-09-08

Last Modified

2026-09-11

Generated

2026-09-29

AI Q&A

2026-09-10

EPSS Evaluated

2026-09-27

NVD

CVE-2026-82537

Affected Vendors & Products

Vendor	Product	Version / Range
roo-code	roo-code	to 3.54.0 (inc)

Helpful Resources

Exploitability

CWE

KEV

CWE ID	Description
CWE-436	Product A handles inputs or steps differently than Product B, which causes A to perform incorrect actions based on its perception of B's state.

Attack-Flow Graph

Executive Summary

Roo-Code through 3.54.0 has a vulnerability where attackers can bypass the auto-approval system by exploiting a mismatch between how the approval gate parses shell commands and how bash interprets them. By inserting a hash character after an allowlisted word, attackers can trick the system into approving a command while bash executes a denied command with elevated privileges on the developer's machine.

Detection Guidance

Detecting this vulnerability requires checking for unauthorized shell command execution in Roo-Code instances. Review logs for commands containing allowlisted words followed by a hash character and denied commands. Inspect shell parser outputs for mismatches between approval gate and bash execution.

Impact Analysis

This vulnerability allows attackers to execute arbitrary shell commands on your machine with the privileges of the Roo-Code agent. This could lead to unauthorized access, data theft, or system compromise if the agent has elevated permissions.

Mitigation Strategies

Update Roo-Code to the latest version beyond 3.54.0. Implement stricter input validation to prevent word-boundary bypasses. Disable auto-execute privileges for shell commands. Monitor logs for suspicious command patterns involving hash characters or separator sequences.

Hi! I’m here to help you understand CVE-2026-82537. Ask me anything about the vulnerability, its impact, or mitigation strategies.

0/70