Grist CVE-2026-55659

Exploitation of the primary vector requires the attacker to already hold document-editor access to a shared Grist document (PR:L), and requires a higher-privileged user - such as the document owner - to subsequently open that document (UI:R), which is why owner escalation is possible only when such a victim views the malicious name/description. The OAuth2-page reflected vector requires a victim to load a URL whose openerOrigin parameter the attacker controls. Limiting factors: exploitation is not possible against a single-user or trusted-only-editor deployment, depends on victim interaction, and is rated high complexity (AC:H) because it hinges on a suitably privileged user loading the payload; it does not affect availability.

The provided CVSS 3.1 vector (AV:N/AC:H/PR:L/UI:R/S:C/C:H/I:H/A:N, base 7.7 High) captures the real profile well: network-reachable, but requiring low-privilege authentication (a document editor, PR:L), user interaction (a victim must open the malicious document, UI:R), and rated high attack complexity because a suitably privileged victim must load the payload. The scope change (S:C) and C:H/I:H reflect the core danger - script executing in the victim's origin crosses a privilege boundary, enabling editor-to-owner escalation, data theft, and modification of access rules; availability is unaffected (A:N). No KEV listing and no POC are indicated in the intelligence, so this is a real but not urgently-exploited priority: the barrier is having write/editor access to a shared document plus a higher-privileged viewer. EPSS is not provided in the input and cannot be cited. The AC:H and UI:R factors meaningfully limit mass exploitation, making this most relevant to multi-tenant or collaboratively-shared Grist deployments rather than internet-wide opportunistic attacks.

An attacker with document-editor rights sets a document's name or description to an HTML/JavaScript payload; when a document owner or other user opens that document, the script runs in their authenticated Grist session and silently rewrites access rules to grant the attacker owner-level control. Alternatively, an attacker crafts a link to the OAuth2 end-of-flow page with a malicious openerOrigin value and lures a victim into clicking it to reflect and execute script. No public exploit code is identified, but the flaw is straightforward stored/reflected XSS.