OpenClaw Feishu CVE-2026-100540

Requires a multi-account OpenClaw Feishu configuration in which the operator has disabled the default account but that account's Feishu credentials remain stored/valid. The attacker must already hold low-privilege access sufficient to invoke the agent's model tool operations (PR:L), and exploitation is gated by the compound precondition reflected in AC:H/AT:P - a single-account or credential-purged deployment is not affected. No end-user interaction is required (UI:N). Limiting factors: this is a non-default operational state (disabled-yet-credentialed default account) and requires pre-existing authenticated access to the OpenClaw agent, so it is not exploitable by an anonymous remote attacker against a default install.

This is a genuine but conditional authorization flaw (CWE-863), not a high-CVSS-but-hollow finding. The vendor CVSS4 vector (7.6, AV:N/AC:H/AT:P/PR:L/UI:N/VC:H/VI:H/VA:N) is internally consistent with the description: network-reachable, but with high complexity (AC:H) and an attack requirement (AT:P) reflecting the specific precondition - a multi-account OpenClaw Feishu deployment where a disabled default account still retains valid credentials. PR:L confirms the actor already holds some legitimate access to the OpenClaw agent/Gateway; this is not an unauthenticated internet-facing bug. No EPSS score, no CISA KEV listing, and no public POC are present in the input, so there is no public exploit identified at time of analysis and no evidence of active exploitation. The realistic priority is moderate: impact is high on confidentiality and integrity of Feishu resources (read/modify through a revoked identity), but exploitation is gated behind a specific non-default multi-account configuration and prior low-privilege access, which is why AC:H/AT:P is appropriate. My independent 3.1 scoring (~6.5) is slightly lower than the vendor's 4.0 rating, driven mainly by VA:N (no availability impact) and the compound configuration prerequisite. No signals conflict; the main gap is the absence of exploitation-probability telemetry (EPSS).

In an OpenClaw deployment integrated with Feishu using multiple accounts, an operator disables the default account (e.g., to revoke a former employee or compromised identity) but its stored credentials are not purged. A low-privileged actor who can drive the agent's model tool operations triggers Feishu resource actions; because the code never checks the disabled flag before selecting the default account, the operations execute under the revoked identity, allowing reads or modifications the actor should no longer be able to perform. No public exploit code is identified at time of analysis.