Privacy policy
Effective October 7, 2026
ourdoc connects development documents with approved code observations. This policy explains account data, collection and model processing.
Google account data
When you choose Google sign-in, we request your basic identity through the openid, email, and profile scopes. We use your verified email address and Google account identifier to create or find your account. We store your name and, when provided, your Google profile picture URL so collaborators can identify your work. We do not store your Google access token, and we do not use Google account data for advertising.
Workspace data
We store the documents, revisions, comments, tasks, invitations, and memberships that people create in the service. Activity records identify who made a change. Members of a workspace can see its content according to their access role. Workspace owners can see member email addresses and manage access. A connected coding agent can access the workspace within the permissions of its token.
Agent read and write metadata, message fetch receipts and reviewed versions remain with their documents to show workflow status. Transfer records support safe retries. Moving a personal document to a team exposes its history and discussions to that team. A team's first completed workflow date may remain while the team exists.
Code observations and consent
Approved collection stores relative file paths, Git changes, content hashes and snapshot times. It also stores checkout identities, session reports, document references and reported checks. Bounded code patches require separate permission. Unsaved editor buffers and full agent conversations are not collected.
Project policy and device approval control collection and external processing. Team roles apply to every project in the team. Personal observations remain private. Revoking a grant stops new collection; pausing or disconnecting does not erase previously stored evidence.
Sessions and hosting
The website uses an HTTP-only cookie to keep your Google sign-in active across tabs and browser restarts for 30 days, unless you sign out sooner. Session credentials are not stored in browser JavaScript storage. The server stores token hashes. Agent tokens expire after 90 days unless revoked sooner. A short-lived cookie protects the Google sign-in callback.
Vercel hosts the website, Google handles account authentication, and a VPS hosts the application API and database. These providers may process connection and request information needed to deliver the service. We do not sell workspace or Google account data.
Product measurement and operations
Optional first-party analytics use a random browser identifier and HTTP-only cookie lasting 180 days. They record known public paths, campaign labels and referrer hostnames, excluding full URLs, query strings, credentials and document text. Do Not Track or Global Privacy Control disables these browser events.
Signed-in operation records cover document and agent activity, code connection, first observations and visualization usage. They contain identifiers, timestamps, versions and operation results, not document or code text. Authorized administrators use them for onboarding, retention and service operations. These records are separate from optional browser analytics.
Raw product events and visitor records last up to 180 days. Aggregated API records last up to 90 days. Signup and first activation dates, with initial source and campaign labels, may remain while the account exists.
Friendly AI processing
Friendly AI sends the saved document title and body to OpenAI with language and formatting instructions. This does not change the source. Explanations are temporarily cached. AI usage records contain outcomes, duration, token counts and estimated costs, without document or code text, for up to 365 days.
Generated development views
With project permission and device approval, HTML generation sends selected code observations and approved patches to OpenAI. Selected documents, session reports and snapshot checks may also be included. The server uses gpt-6.1-sol with medium reasoning. Generated HTML is isolated from your session credentials.
We request store: false for OpenAI calls. This is not a promise of zero provider retention or a complete description of provider data handling. Generation results and their caches follow the underlying code evidence lifetime.
Retention and requests
Code evidence defaults to 7 days of retention, configurable from 1 to 30 days. Delete collected data removes live source evidence and derived views. Expired or deleted evidence cannot support a current verified result. Existing database backups rotate after 14 days; document history has a separate lifetime.
Workspace records remain until they are removed by an administrator. Archiving hides an item from the active workspace but does not erase it. Database backups rotate after about fourteen days. To request access, correction, or deletion of your account data, email us at peerproblem@gmail.com.
Changes
We will update this page when data practices change. The effective date above shows when this version took effect.