CVE-2026-26321: OpenClaw path traversal enables local

CISO Take

CVE-2026-26321 is a path traversal flaw (CWE-22) in OpenClaw's Feishu extension where the `sendMediaFeishu` tool accepts attacker-controlled `mediaUrl` values as raw filesystem paths, enabling reads of arbitrary local files such as SSH keys, `.env` secrets, or `/etc/passwd` with zero authentication. The CVSS 7.5 score (AV:N/AC:L/PR:N/UI:N/C:H) reflects a frictionless exploitation profile — any actor able to influence OpenClaw's tool calls, whether directly or via prompt injection against ingested external content, can silently exfiltrate sensitive files. While there is no CISA KEV entry or public exploit yet, OpenClaw carries 11 total CVEs and AIID #1368 documents active abuse of its extension ecosystem for credential theft, indicating this package and its plugin surface are under active adversarial attention. Patch to OpenClaw 2026.2.14 immediately; if patching is not feasible, disable the Feishu extension and sandbox the assistant process with filesystem restrictions.

Sources: NVD GitHub Advisory ATLAS

What is the risk?

High risk for any deployment running OpenClaw prior to 2026.2.14 with the Feishu extension enabled. The network-accessible, zero-privilege, no-user-interaction CVSS profile means the attack requires no special position beyond the ability to feed crafted input into the agent's context. Personal AI assistants typically run with broad local filesystem access, giving a successful exploit direct reach to credentials, config files, and secrets. The 11-CVE track record and documented real-world abuse of OpenClaw's extension ecosystem (AIID #1368) elevates practical risk beyond what CVSS alone conveys.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
OpenClaw	pip	—	No patch
4 dependents 41% patched ~3d to patch Full package profile →

Do you use OpenClaw? You're affected.

How severe is it?

CVSS 3.1

7.5 / 10

EPSS

0.5%

chance of exploitation in 30 days

Higher than 38% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

No known exploitation

Sophistication

Trivial

What is the attack surface?

AV Network

AC Low

PR None

UI None

S Unchanged

C High

I None

A None

What should I do?

6 steps

Patch immediately: upgrade to OpenClaw 2026.2.14 (fix commit 5b4121d6, GHSA-8jpq-5h99-ff5r).
If patching is not immediately possible, disable the Feishu extension in OpenClaw settings.
Sandbox the OpenClaw process using OS-level controls (chroot, container, or macOS sandbox profiles) to restrict filesystem access to necessary directories only.
Audit Feishu outbound API call logs for unexpected payloads containing file content.
Review all AI agent tool definitions for parameters that accept file paths and enforce allow-list validation.
Monitor for prompt injection patterns in content ingested by the assistant (emails, documents, web pages).

What does CISA's SSVC say?

Decision Track

Exploitation none

Automatable Yes

Technical Impact partial

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Data Extraction Prompt Injection Privacy Violation Agent Plugin AML.T0037 - Data from Local System AML.T0051.000 - Direct AML.T0051.001 - Indirect AML.T0053 - AI Agent Tool Invocation AML.T0086 - Exfiltration via AI Agent Tool Invocation

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15 - Accuracy, robustness and cybersecurity

ISO 42001

A.6.2.3 - AI system security controls

NIST AI RMF

MANAGE 2.2 - Manage AI risks through established controls

OWASP LLM Top 10

LLM07 - Insecure Plugin Design

Frequently Asked Questions

What is CVE-2026-26321?

CVE-2026-26321 is a path traversal flaw (CWE-22) in OpenClaw's Feishu extension where the `sendMediaFeishu` tool accepts attacker-controlled `mediaUrl` values as raw filesystem paths, enabling reads of arbitrary local files such as SSH keys, `.env` secrets, or `/etc/passwd` with zero authentication. The CVSS 7.5 score (AV:N/AC:L/PR:N/UI:N/C:H) reflects a frictionless exploitation profile — any actor able to influence OpenClaw's tool calls, whether directly or via prompt injection against ingested external content, can silently exfiltrate sensitive files. While there is no CISA KEV entry or public exploit yet, OpenClaw carries 11 total CVEs and AIID #1368 documents active abuse of its extension ecosystem for credential theft, indicating this package and its plugin surface are under active adversarial attention. Patch to OpenClaw 2026.2.14 immediately; if patching is not feasible, disable the Feishu extension and sandbox the assistant process with filesystem restrictions.

Is CVE-2026-26321 actively exploited?

No confirmed active exploitation of CVE-2026-26321 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-26321?

1. Patch immediately: upgrade to OpenClaw 2026.2.14 (fix commit 5b4121d6, GHSA-8jpq-5h99-ff5r). 2. If patching is not immediately possible, disable the Feishu extension in OpenClaw settings. 3. Sandbox the OpenClaw process using OS-level controls (chroot, container, or macOS sandbox profiles) to restrict filesystem access to necessary directories only. 4. Audit Feishu outbound API call logs for unexpected payloads containing file content. 5. Review all AI agent tool definitions for parameters that accept file paths and enforce allow-list validation. 6. Monitor for prompt injection patterns in content ingested by the assistant (emails, documents, web pages).

What systems are affected by CVE-2026-26321?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, personal AI assistants, agentic tool execution.

What is the CVSS score for CVE-2026-26321?

CVE-2026-26321 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.48%.

What is the AI security impact?

Affected AI Architectures

agent frameworkspersonal AI assistantsagentic tool execution

MITRE ATLAS Techniques

AML.T0037 Data from Local System

AML.T0051.000 Direct

AML.T0051.001 Indirect

AML.T0053 AI Agent Tool Invocation

AML.T0086 Exfiltration via AI Agent Tool Invocation

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: A.6.2.3

NIST AI RMF: MANAGE 2.2

OWASP LLM Top 10: LLM07

What are the technical details?

Original Advisory

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 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.

Exploitation Scenario

An attacker embeds a prompt injection in a document or email that OpenClaw is asked to process: the injected instruction directs the assistant to call `sendMediaFeishu` with `mediaUrl` set to `/home/user/.ssh/id_rsa`. The vulnerable extension reads the private key from disk and transmits it as a media payload to the attacker's Feishu contact, with no authentication, no user confirmation prompt, and no error raised. In a more automated variant, the attacker publishes a malicious webpage or Feishu message containing the injection, which triggers silently when OpenClaw browses or processes it as part of an agentic task, exfiltrating credentials to a controlled endpoint before the user is aware.

Weaknesses (CWE)

CWE-22 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') Primary

CWE-22 — Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal'): The product uses external input to construct a pathname that is intended to identify a file or directory that is located underneath a restricted parent directory, but the product does not properly neutralize special elements within the pathname that can cause the pathname to resolve to a location that is outside of the restricted directory.

[Implementation] Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does. When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue." Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylis
[Architecture and Design] For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server.

Source: MITRE CWE corpus.