CVE-2026-70558Dinky's POST /download/uploadFromRsByLocal handler passes the caller-supplied path parameter directly to new File(path) and file.transferTo(dest) with no path validation. The route is marked @SaIgnore and /download/** is excluded from the Sa-Token interceptor, so the only guard is a header equality check against a dinkyToken value whose default (efda1551-7958-4e0f-80a8-dfd107df3e38) is hardcoded in source and shipped to every deployment. Anyone who can reach Dinky's HTTP port (8888 by default) and supplies the hardcoded token can write arbitrary files as the Dinky service account. The default Docker image runs on 8888 with no proxy or authentication and chmod 777 on /opt/dinky, so the application's own classpath, launch scripts, and static assets are writable. Demonstrated impact: overwriting /opt/dinky/config/static/index.html served attacker JavaScript to admin browsers immediately, and writing /opt/dinky/org/dinky/Dinky.class executed attacker code as the Dinky service account at the next JVM start via a classpath-shadow launched by script/bin/auto.sh. Writes are uid 9999 (flink), not root, so /etc, /root, /home, and /usr are refused. Affects Dinky v1.2.5 (the current release) and the development branch, where the code is byte-identical.
2026-08-06 · score 9.8
CVE-2026-67622Flowise through 3.1.4 contains an insecure direct object reference vulnerability in the OpenAI Assistants integration that allows authenticated attackers to access credentials belonging to other workspaces by supplying an arbitrary credential UUID to Assistants endpoints without workspace ownership verification. Attackers can enumerate cross-workspace assistant metadata, retrieve file and vector store listings, and upload files into victim workspaces by exploiting the missing workspace-scoped authorization check in the credential lookup logic.
2026-08-06 · score 9.9
CVE-2026-48088OpenReception's appointment booking software provides an end-to-end encrypted appointment booking platform. Prior to version 1.0.4, the route `POST /api/tenants/{tenantId}/staff/{staffId}/crypto` accepts and stores attacker-controlled ML-KEM-768 public keys against any tenant on the platform without authentication. The handler logs an "Unauthorized crypto key storage attempt" warning when neither a session nor a registration cookie is present, then proceeds to insert the row regardless. The platform's E2E claim that "even administrators cannot view sensitive information" is broken: any unauthenticated network attacker can register themselves as an additional encryption recipient for any tenant's future patient appointments. A second variant of the bug suppresses the unauthorized-warning log entry. The Zod schema makes the `email` field optional. When the request body omits `email` and the request carries no registration cookie, the comparison `registrationEmail === email` becomes `undefined === undefined`, which evaluates to `true`. The handler treats the request as a legitimate registration flow, skips the warning entirely, and stores the row. Successful storage is still recorded as an `[info]` log line, but the security-relevant warning that operators are most likely to monitor or alert on is gone. The `staff_crypto` table has no unique constraint on `user_id`, so an arbitrary number of attacker rows can coexist for the same staff identifier and all return as `is_active=true`. The supplied `staffId` does not need to match any existing user or pending invite. Schema validation on `passkeyId`, `publicKey`, and `privateKeyShare` is also weak: the literal string `<placeholder-base64>` was accepted, indicating no length, format, or cryptographic-validity check beyond field presence. This weakness is independent of the auth bypass but compounds it: a poisoned directory can also be filled with malformed entries that break legitimate booking flows. The injected key is consumed by the public booking flow. After completing the unauthenticated `bootstrap-challenge` and `bootstrap-verify` ceremony as a "patient", the resulting `bookingAccessToken` is accepted by `GET /api/tenants/{id}/appointments/staff-public-keys`, which returns the attacker-controlled keys alongside any legitimate ones. A new appointment encrypts its tunnel key with ML-KEM to all listed recipients, so the attacker becomes a co-recipient of the encryption and can decapsulate the tunnel key with the matching secret. From there, all appointment payloads for that booking are decryptable. Version 1.0.4 patches the issue.
2026-08-06 · score 9.4
CVE-2026-48087OpenReception's appointment booking software provides an end-to-end encrypted appointment booking platform. Prior to version 1.0.2, the registration handler at `POST /api/auth/register/{userId}` validates the relationship between the WebAuthn challenge and the registration cookie's email but never validates that the `userId` in the URL belongs to that email. An unauthenticated attacker requests a challenge for their own email, generates a registration response with their own authenticator, and submits it against any victim user's URL. The challenge-vs-cookie email match passes, the WebAuthn ceremony validates, and `addPasskey` writes the attacker's credential into the victim's `user_passkey` rows. The next victim-email login accepts a passkey assertion from the attacker's authenticator and issues a session as the victim. User IDs are not strictly secret on this platform, but the exact set of exposure surfaces should be assessed by the maintainers. Staff-list endpoints return user IDs to authenticated tenant members per the route signature; live verification of all exposure surfaces (whether user IDs leak through any unauthenticated route, through invite-confirmation URLs, or through other administrative views) is part of the pending live PoC. Where the attacker knows the victim's email and userId, the analysis below becomes account takeover. Version 1.0.2 fixes the issue.
2026-08-06 · score 9.8
CVE-2026-48086OpenReception's appointment booking software provides an end-to-end encrypted appointment booking platform. Prior to version 1.0.2, a TENANT_ADMIN promotes themselves to platform-wide GLOBAL_ADMIN through a single PUT request. The role-update handler accepts the `GLOBAL_ADMIN` enum value from any tenant admin updating their own tenant's staff. No policy check enforces that "only an existing GLOBAL_ADMIN may grant GLOBAL_ADMIN", so the schema validation IS the authorization decision. After re-login, the JWT contains the new role and the formerly-tenant-scoped admin reaches every other tenant on the platform. On the hosted OpenReception service this is a scope-changed escalation: a single customer-side tenant administrator gains full platform-wide administrative control over all other tenants' configuration, users, staff records, operational metadata, and tenant lifecycle. Plaintext appointment contents remain subject to the E2E model unless chained with the staff-crypto poisoning issue (V-4) or with staff-passkey hijacking (V-1). On a single-tenant self-hosted deployment it is still a privilege escalation because TENANT_ADMIN should not be able to create new tenants, modify global configuration, or manage other administrators. The same handler also accepts updates targeted at any colleague within the tenant. A tenant admin can promote a separate collaborator account instead of themselves, leaving their own audit trail clean while the platform-wide breach happens through a separate identity. Version 1.0.2 fixes the issue.
2026-08-06 · score 9.9
