CVE-2026-26321
Published: 19 February 2026
Summary
CVE-2026-26321 is a high-severity Path Traversal (CWE-22) vulnerability in Openclaw Openclaw. Its CVSS base score is 7.5 (High).
Operationally, exploitation aligns with the MITRE ATT&CK technique Data from Local System (T1005); ranked at the 8.2th percentile by exploit likelihood (below the median); it is not currently listed in the CISA KEV catalog.
This vulnerability is AI-related — categorised as Other AI Platforms; in the LLM/Generative AI Risks risk domain.
Threat & Defense at a Glance
Threat & Defense Details
Likely Mitigating ControlsAI
Per-CVE control mapping for this CVE has not run yet; the list below is derived from the weakness types (CWEs) cited in the NVD entry.
Validates pathnames and filenames to prevent traversal outside intended directories.
MITRE ATT&CK Enterprise TechniquesAI
Why these techniques?
Path traversal in sendMediaFeishu directly allows arbitrary local file reads (e.g. /etc/passwd), enabling collection of data from the local filesystem without requiring additional access primitives.
NVD Description
OpenClaw is a personal AI assistant. Prior to OpenClaw version 2026.2.14, the Feishu extension previously allowed `sendMediaFeishu` to treat attacker-controlled `mediaUrl` values as local filesystem paths and read them directly. If an attacker can influence tool calls (directly or via…
more
prompt injection), they may be able to exfiltrate local files by supplying paths such as `/etc/passwd` as `mediaUrl`. Upgrade to OpenClaw `2026.2.14` or newer to receive a fix. The fix removes direct local file reads from this path and routes media loading through hardened helpers that enforce local-root restrictions.
Deeper analysisAI
CVE-2026-26321 is a path traversal vulnerability (CWE-22) in the Feishu extension of OpenClaw, a personal AI assistant, affecting versions prior to 2026.2.14. The `sendMediaFeishu` function improperly treats attacker-controlled `mediaUrl` values as local filesystem paths, allowing direct reading of arbitrary files on the host system.
Attackers can exploit this vulnerability remotely without privileges (AV:N/AC:L/PR:N) if they can influence tool calls, either directly or via prompt injection techniques common in AI assistants. By supplying a sensitive local path such as `/etc/passwd` as the `mediaUrl`, they can exfiltrate file contents, achieving high confidentiality impact as reflected in the CVSS v3.1 score of 7.5 (C:H/I:N/A:N/S:U).
The official mitigation is to upgrade to OpenClaw version 2026.2.14 or newer, which eliminates direct local file reads in this function and redirects media loading through hardened helpers that enforce local-root restrictions. Details are provided in the OpenClaw GitHub security advisory (GHSA-8jpq-5h99-ff5r), the release notes for v2026.2.14, and the fix commit (5b4121d6011a48c71e747e3c18197f180b872c5d).
This issue underscores risks in AI assistant tool extensions, where prompt injection can lead to local file disclosure. No real-world exploitation has been reported as of the CVE publication on 2026-02-19.
Details
- CWE(s)
Affected Products
AI Security AnalysisAI
- AI Category
- Other AI Platforms
- Risk Domain
- LLM/Generative AI Risks
- OWASP Top 10 for LLMs 2025
- None mapped
- Classification Reason
- Matched keywords: ai, prompt injection