Google sign-in
Shado uses Clerk to provide authentication. When you choose Google, Clerk receives basic Google identity information, including your Google account identifier, email address and verification status, and profile information such as your name and profile photo. This information is used to sign you in and maintain your account and profile.
Google sign-in requests basic identity permissions. It does not request access to your Gmail messages or Google Drive files.
Your account and connected sign-in methods
Your profile and linked sign-in methods are maintained with your Clerk account. Shado uses account identifiers to associate workspace access, projects, and credit records. Account details such as your name and email may also appear to workspace managers in member and activity views.
Connecting Google adds a sign-in method to your existing Shado account. It keeps the same account identity, projects, and credits; it does not merge two existing Shado accounts.
You can manage your Google connection in Account → Security. You can disconnect Google when another usable sign-in method remains. Disconnecting Google does not delete your Shado account, projects, or credit records.
Service and security records
Shado uses account, request, billing, and security records to operate the service and document agreement acceptance. Content necessary for a feature may be sent to the providers involved in that feature. See the Terms of Service for the existing terms about content, providers, and storage.
When browser security collection is enabled, signed-in browsers provide a browser identifier and an automation indicator to help investigate account and promotion abuse. Shado stores keyed hashes of the identifier, session, and platform-observed network address; raw fingerprint components are not submitted, and this collection sends no data to FingerprintJS. Voluntary automation reports contain only the agent name, operator contact, and purpose the sender chooses to disclose; those declarations are unverified. Security records expire after 30 days and are removed by scheduled cleanup. These signals alone do not establish a person's identity or automatically suspend an account.
When free usage protection is enabled, Shado also uses your verified account and sign-in provider's browser client, together with hashed email and platform-observed network information, to limit repeated free allowance claims. This may temporarily hold new free usage while preserving purchased credit and accepted work. Account bindings, monthly claims, and activation baselines expire after 90 days; expired operator overrides are removed after 90 days and control audit records after 180 days by scheduled cleanup. Policy configuration and financial records follow their separate service retention requirements. Account deletion does not immediately erase these limited abuse prevention records.
When abuse risk observation is enabled, Shado records counts of verified account associations and free-usage admission attempts to support abuse review and model evaluation. Missing evidence is recorded as unavailable. Risk estimates alone do not suspend accounts or change billing. Observation evidence expires after 90 days and human review, coverage, and audit records after 180 days through scheduled cleanup; registered model artifacts and configuration are retained for versioning. Private review and calibration exports follow the service's operator access controls.
Signup tracking records verified account creation time, email verification method, and available invitation and referral evidence. These operational records are kept separately from browser-reported acquisition data. Signup records and queued analytics events become eligible for removal after 90 days through bounded scheduled cleanup.
Landing page and signup analytics
We use PostHog on the public landing page to understand section visits, plan and showcase selections, and clicks that start sign-in or submit an idea. Events contain fixed interaction labels, the interface language, and a random identifier stored for the browser session. Landing interaction properties exclude prompt text, comments, uploaded files, account details, page URLs and referrers. Session recordings and automatic interaction capture are disabled.
To understand how new accounts find Shado, we retain a browser-reported first landing origin and public entry path, referring domain, tagged campaign labels, and random session identifier. The bounded visit identifier and source are carried through Shado sign-in links and connected to the verified account identifier after signup. Private page paths, URL fragments, unrelated query parameters, names, email addresses, and canvas contents are excluded. When PostHog is configured, permitted signup observations and subsequent invitation/referral evidence updates are delivered there; signed provider events alone do not send an account to PostHog.
This integration respects browser Do Not Track and Global Privacy Control signals. Analytics is disabled on local development and preview pages.
In-app usage analytics
To learn where Shado works well and where people get stuck, we record a fixed list of in-app events linked to your account identifier: creating and opening canvases, adding nodes by type, uploads by file type and size range, generation requests with the model used and the outcome, agent and voice messages, downloads, plan views, checkouts and subscription changes, referral and sharing actions, and free usage limits. Events are sent to PostHog.
Events never include prompts, agent conversations, file names, canvas names or contents, generated media, email addresses or IP addresses. Session recordings and automatic interaction capture are not used.
If your browser sends Do Not Track or Global Privacy Control while you are signed in, we stop recording these events for your account, including on other devices.
Questions and requests
Contact hello@ghostframes.ai about privacy or deletion requests.