Typebot CVE-2026-47704

Exploitation requires the attacker to hold a valid authenticated session on the Typebot instance with at least read-level access to one typebot, confirmed by the CVSS PR:L metric. The attacker must additionally know or be able to enumerate a live resultId for a webhook session currently in a suspended/waiting state in a different typebot; resultIds are typically UUIDs and are not publicly advertised, requiring some enumeration capability or insider knowledge of the target instance. The vulnerability is specific to the webhook block feature: typebots that do not use webhook blocks with waiting/suspended sessions are not reachable via this attack path. No non-default configuration is required beyond the webhook feature being in use - any shared or multi-user instance running a vulnerable version with active webhook sessions is exposed.

The CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:N with VI:H) reflects that this vulnerability is network-reachable, requires no special attack conditions, demands only low-privilege authentication, and produces a high integrity impact - an accurate characterisation of an IDOR flaw that any authenticated platform user can trigger without sophisticated tooling. No CISA KEV listing exists and no public POC code has been identified, meaning widespread opportunistic exploitation is not yet confirmed, though the low technical barrier makes exploitation realistic for any motivated insider or malicious co-tenant on a shared instance. The absence of EPSS data in the available intelligence means exploitation probability cannot be quantified, but the low complexity aligns with SSVC's automatable signal - an attacker could enumerate or observe live resultIds and systematically tamper with sessions at scale. Even a mid-sized organisation running a shared instance for customer intake workflows - such as a municipal services team in kielce - should treat this as an urgent upgrade given that a single rogue authenticated user can corrupt the webhook-driven flows of every other typebot on the deployment. The clear availability of a vendor-confirmed patch in v3.17.0 and the absence of a scope change (impact remains within the vulnerable system) keep this from escalating further, but the integrity damage - injecting arbitrary JSON to advance business logic - is a concrete, actionable threat rather than a theoretical one.

An authenticated attacker with read-only access to their own typebot identifies or enumerates the live resultId of a suspended webhook session belonging to a victim typebot on the same instance - for example by monitoring application responses or using knowledge gained as a platform co-tenant. The attacker crafts a webhook resume request that supplies their own authorized typebotId and a valid blockId alongside the victim's resultId; the handler passes the authorization check against the attacker's typebot, resolves the victim's result record, and advances the victim session with arbitrary injected JSON, potentially corrupting collected data or triggering unintended downstream automations in the victim's workflow.