ClawHub’s moderation mechanism allows four ordinary authenticated users to report the same visible skill, which automatically hides it from the catalog without any moderator approval. The flaw permits an attacker to remove legitimate skills or hide official content from all consumers, effectively tampering with the platform’s integrity and availability for skill discovery.

The vulnerability affects self‑hosted deployments of the openclaw:clawhub application backend. All revisions before 8c2de6c506 are potentially impacted; the exact historical range is not fully known, though the issue was confirmed at revision cbfee7343ddc867316dd9b3de6fa8856730f9f41. The npm CLI and OpenClaw runtime are not affected.

The CVSS score of 8.7 classifies this as a high‑severity flaw. Although EPSS is not available and it is not in CISA KEV, the attack requires coordination of four authenticated accounts and does not rely on elevated privileges. If an attacker can control or create four such accounts, the skill hiding can be achieved. The lack of a moderator decision in the flow makes the vulnerability simpler to exploit once the quorum is met.