CVE WatchAug 18, 2026

CVE-2026-75858

CodeWhale (packages codewhale / codewhale-tui) versions >= 0.8.41 and < 0.8.64 contain a remote code execution vulnerability in the rlm_eval

CVSS

7.8

High

EPSS

0.3%

p19

KEV

—

Exploit Today

0-100

Published: Aug 18, 2026 · Last modified: Aug 18, 2026 · CWE-94

EPSS · 30d

0.3%EPSS · 30 days0.3%

2026-08-192026-08-31

Technical description

CodeWhale (packages codewhale / codewhale-tui) versions >= 0.8.41 and < 0.8.64 contain a remote code execution vulnerability in the rlm_eval tool. The tool's approval_requirement() returns ApprovalRequirement::Auto, which the engine treats as 'never prompt,' causing arbitrary model-supplied Python code to run in a python3 interpreter without consulting the user's configured --approval-policy and without any approval prompt or audit step. An attacker can induce the agent to execute arbitrary code via prompt injection in untrusted content the agent reads (a web page, fetched URL, repository file, or MCP tool result); the companion rlm_open tool can stage such content. Code runs on the user's machine at the user's privilege level. Fixed in 0.8.64.

Official references

Related CVEs

CVECVSSEPSSKEVRExploitTitleMod.

CVE-2026-585728.8 HIG

—

———Dell PowerStore contains a Code Injection vulnerability. An authenticated user with limited privileges could potentially exploit this vulnerability to execute arbitrary code with root privileges.7h CVE-2026-188089.8 CRI

—

———Improper Control of Generation of Code ('Code Injection') vulnerability in Klemsan Electrical Electronics Inc. KIO (Klemsan Internet Objects) allows Code Injection. This issue affects KIO (Klemsan Internet Objects): before v1.9.12h CVE-2026-4813—

—

———A vulnerability in the Lutece Core XSL export management module up to version 7.1.7, which allows authenticated administrators to execute code remotely. The XML/XSLT processing configuration does not enable secure processing mode (FEATURE_SECURE_PROCESSING), allowing Java extension functions to be executed from malicious XSL stylesheets. An attacker with administrator privileges can upload a manipulated XSL transformation file and trigger its execution during user export operations, resulting in the execution of arbitrary code on the server.8h CVE-2026-823937.5 HIG

—

———pnpm is a package manager. Prior to 10.34.5 and 11.11.0, pnpm accepts a scoped path traversal in a tarball dependency's package.json manifest name because pnpm11/resolving/npm-resolver/src/pickPackage.ts rejects slash characters only for unscoped names. During pnpm install, the unvalidated name reaches raw path joins in pnpm11/installing/deps-resolver/src/resolvePeers.ts, pnpm11/installing/deps-resolver/src/index.ts, and pnpm11/deps/graph-builder/src/lockfileToDepGraph.ts, causing package extraction outside node_modules and allowing attacker-controlled files to overwrite arbitrary filesystem paths even when --ignore-scripts is used. The overwrite can replace shell startup files, Git hooks, or installed package code and lead to code execution. This issue is fixed in versions 10.34.5, and 11.11.0.13h CVE-2026-828054.3 MED

—

——0A vulnerability was found in Typora up to 1.13.8/1.14.6. This vulnerability affects unknown code of the component Mermaid Rendering Engine. The manipulation of the argument classDef/style results in cross site scripting. The attack may be launched remotely. The exploit has been made public and could be used. Upgrading to version 1.14.8 is able to resolve this issue. You should upgrade the affected component. The vendor was contacted early, responded in a very professional manner and quickly released a fixed version of the affected product.2d CVE-2026-827004.3 MED

—

——0A vulnerability was found in code-projects Online Shopping System 1.0. Affected by this vulnerability is an unknown functionality of the file /offersmail.php of the component Newsletter Subscription. The manipulation of the argument email results in cross site scripting. The attack may be performed from remote. The exploit has been made public and could be used.2d