CVE-2026-27484: 
Homebrew vulnerability analysis and mitigation

CVE-2026-27484 is a missing authorization vulnerability in OpenClaw, a personal AI assistant, affecting its Discord moderation action handling (timeout, kick, ban). In versions 2026.2.17 and below, the bot uses sender identity from untrusted request parameters in tool-driven flows rather than from trusted runtime sender context, allowing a non-admin user to spoof sender identity fields and trigger unauthorized moderation actions. The vulnerability was published on February 20, 2026, and patched in version 2026.2.18. It carries a CVSS v3.1 base score of 4.3 (Medium) and a CVSS v4.0 base score of 2.3 (Low) (GitHub Advisory, Security Advisory).

The root cause is CWE-862 (Missing Authorization): the Discord moderation handler (discord-actions-moderation.ts) read the senderUserId field directly from tool action parameters rather than from the trusted runtime context established at session initialization. In tool-driven flows, this meant any authenticated (low-privilege) user could craft a request supplying an arbitrary senderUserId value, effectively impersonating an admin or moderator. The fix introduced a verifySenderModerationPermission function that calls hasGuildPermissionDiscord to validate the requester's actual guild permissions (e.g., BAN_MEMBERS, KICK_MEMBERS, MODERATE_MEMBERS) before executing any moderation action, and propagates the trusted requesterSenderId from inbound session context rather than tool arguments (Fix Commit, GitHub Advisory).

Successful exploitation allows a non-admin Discord user to perform unauthorized moderation actions — including timing out, kicking, or banning other guild members — in any Discord server where the OpenClaw bot is deployed with the necessary guild permissions and moderation actions are enabled. This compromises the integrity of community management, potentially enabling malicious users to disrupt or remove legitimate members from Discord guilds. There is no confidentiality or direct system availability impact; the scope is limited to the integrity of Discord guild membership and moderation state (GitHub Advisory, Feedly).

There is no public proof-of-concept exploit and no evidence of in-the-wild exploitation as of the time of disclosure (Feedly). The vulnerability requires the attacker to have low-level authenticated access to the OpenClaw bot (i.e., be a registered user in the tool-driven flow), and exploitation is only possible when Discord moderation actions are enabled and the bot holds the relevant guild permissions. The EPSS score is approximately 0.03% (10th percentile), indicating a very low probability of exploitation in the near term (GitHub Advisory). The vulnerability is not listed in the CISA KEV catalog and no threat actor attribution has been reported.

1. Identify a target deployment: Locate an OpenClaw instance (npm package openclaw) running version ≤2026.2.17 with Discord moderation actions enabled and the bot holding guild moderator permissions (ban, kick, or timeout).
2. Obtain low-privilege access: Authenticate to the OpenClaw tool-driven flow as any non-admin user — this requires only standard user-level access to the bot's interface.
3. Craft a spoofed moderation request: Submit a tool-driven action request (e.g., ban, kick, or timeout) and include a senderUserId parameter set to the Discord user ID of a legitimate admin or moderator, rather than your own user ID.
4. Trigger unauthorized moderation action: Because the vulnerable code reads senderUserId directly from the request parameters without validating actual guild permissions, the bot executes the moderation action (e.g., banning a target user) as if it were initiated by the spoofed admin identity.
5. Achieve objective: The target Discord guild member is timed out, kicked, or banned without any legitimate admin authorization (Fix Commit, GitHub Advisory).

• Logs: OpenClaw application logs showing moderation actions (timeout, kick, ban) initiated by a senderUserId that does not match the authenticated session user or any known admin; unexpected moderation events in Discord audit logs attributed to the bot without corresponding admin-initiated requests.
• Discord Audit Log: Sudden or unexplained bans, kicks, or timeouts executed by the OpenClaw bot account, especially for users who were not targeted by any known admin action.
• Application Behavior: Tool-driven flow requests containing a senderUserId parameter that differs from the session's authenticated user identity, particularly if the supplied ID corresponds to a guild admin or moderator.

The primary remediation is to upgrade the openclaw npm package to version 2026.2.18 or later, which enforces trusted sender context (requesterSenderId) and validates guild permissions before executing any moderation action (GitHub Advisory, Release Notes). For deployments unable to patch immediately, disable Discord moderation actions in the OpenClaw configuration, or revoke the bot's guild moderator permissions (ban, kick, timeout) until the update can be applied (Feedly).

The vulnerability was reported by @aether-ai-agent and published by orlyjamie on February 20, 2026, with the fix credited to the same reporter (Security Advisory). No significant broader industry commentary, media coverage, or social media discussion has been identified for this vulnerability.