همهٔ گزارش‌ها
زیاد

OpenClaw: Unsanitized CWD path injection into LLM prompts

متن اصلی

نسخه‌های متأثر

نسخه‌های متأثر
< 2026.2.15
نسخهٔ اصلاح‌شده
2026.2.15

بهره‌برداری

CVE-2026-27001

احتمال بهره‌برداری در ۳۰ روز آینده: 0.22%

FIRST EPSS · ۱۴۰۵/۰۸/۲۹

شرح آسیب‌پذیری

Overview

OpenClaw embedded the current working directory (workspace path) into the agent system prompt without sanitization. If an attacker can cause OpenClaw to run inside a directory whose name contains control/format characters (for example newlines or Unicode bidi/zero-width markers), those characters could break the prompt structure and inject attacker-controlled instructions.

Impact

Prompt injection may alter agent behavior and could lead to unintended tool use or disclosure of sensitive information.

Affected Packages / Versions

  • Package: openclaw (npm)

  • Vulnerable versions: < 2026.2.15 (latest published vulnerable version as of 2026-02-16: 2026.2.14)

  • Patched versions: >= 2026.2.15

Fix

The workspace path is now sanitized before it is embedded into any LLM prompt output, stripping Unicode control/format characters and explicit line/paragraph separators. Workspace path resolution also applies the same sanitization as defense-in-depth.

Fix Commit(s)

  • 6254e96acf16e70ceccc8f9b2abecee44d606f79

Thanks @aether-ai-agent for reporting.

پیوندها

گزارش‌دهندگان

  • aether-ai-agent