Openclaw CVE-2026-22171

OpenClaw versions before 2026.2.19 with Feishu integration enabled. Attacker must control Feishu media key values returned to client (requires network-level MITM, Feishu account compromise, or malicious Feishu server). Remote unauthenticated exploitation; no victim interaction needed.

The CVSS v3.1 score of 8.2 with vector AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:L/A:N indicates a high-severity network-exploitable vulnerability requiring no authentication, low attack complexity, and no user interaction, with high confidentiality impact and low integrity impact. The attack is automatable given the straightforward nature of path traversal exploitation. However, real-world risk is moderated by several factors: the vulnerability is specific to the Feishu integration component which may not be enabled in all deployments, exploitation requires the attacker to control media key values returned to the OpenClaw client (suggesting compromise or manipulation of Feishu integration flows), and file writes are constrained to process permissions. No EPSS score is provided, and the absence of KEV listing or public proof-of-concept suggests exploitation has not been observed in the wild. Using SSVC framework: exploitation is theoretical but plausible given low complexity, technical impact is partial (file write within permission boundaries), and the vulnerability is automatable. This represents a genuine priority for organizations using the Feishu integration, but may be lower priority for deployments not utilizing this specific extension.

An attacker intercepts or manipulates the communication channel between OpenClaw and Feishu services (or compromises Feishu integration credentials) to inject malicious media key values containing path traversal sequences such as '../../../../etc/cron.d/malicious' or '../../../.ssh/authorized_keys'. When OpenClaw processes the media download request, it constructs the file path by concatenating os.tmpdir() with the malicious media key, escaping the temporary directory and writing attacker-controlled content to sensitive system locations. Given the network-accessible attack vector (AV:N) and low complexity (AC:L) with no authentication required (PR:N), exploitation could be automated once an attacker identifies the vulnerable Feishu integration endpoint and can inject crafted media key values into the download flow.