Stride MCP connector — privacy statement

The Stride MCP connector lets an AI assistant (currently Claude Code or Codex, each via its own manually-registered client) look up your own Stride learning-and-growth status on your behalf, with your explicit consent. Since 2026-09-11 it is write-enabled: it can read your own work and status, start an assessment attempt, attach evidence, and submit a provisional score on your behalf. It cannot publish anything, mark your work complete, or pay a reward.

Claude Code and Codex are in live use on pilot. Support for hosted Claude (Claude.ai) is planned: its client is registered there, but Stride has not yet enabled or tested that connection end to end.

Every AI client that connects is manually registered and approved by Stride first — there is no self-service or public sign-up flow today, on any platform. See Connect Stride to your AI assistant for exactly which client is connected today and what's only planned.

Only data about the connecting user's own Stride account, and only within that user's own organization for org-scoped content:

• your connector-grant and consent state (which permissions you've approved, and when they expire);
• a count and summary of your own recent assessment attempts (state, title, timestamps — not the content of any submission);
• eligible weekly challenges and quest steps visible to your organization, including their instructions text (this is content written by a Stride content author, not personal data about you);
• the status of one specific assessment attempt of yours — looked up with an opaque reference handle only you hold, not a raw database ID, so nobody can guess or reuse someone else's handle to see their attempt.

It cannot access another user's data, any admin function, or any write action beyond starting your own assessment attempt, attaching your own evidence, and submitting your own provisional score.

This data is used only to answer the questions you ask the assistant about your own Stride status and available work — nothing else. It is not used for advertising, profiling, or any purpose unrelated to answering your request.

Employment-use boundary. Output from this connector must never be used for hiring, promotion, compensation, discipline, termination, employability, or any other employment decision about you. Any assessment result surfaced through this connector is provisional and non-authoritative.

Two vendors, both already used by Stride generally and not brought in specifically for this connector: Vercel (application hosting) and Supabase (database and authentication). Neither vendor's underlying model or infrastructure is given access to this data beyond running Stride's own code — no third-party AI training occurs on this data.

Placeholder — the production data-residency commitment for a real customer-facing listing has not been decided or documented — Ryan / Privacy-Legal to confirm before this statement is published for a real listing. Today's development environment stores data in a Supabase database provisioned in the us-east-1 AWS region, which is not itself a production residency commitment.

• Your OAuth authorization request and code are short-lived (minutes). Using one, or letting it expire, immediately makes it permanently unusable for authorization. Physical deletion of the row is now scheduled daily; its first live run is not yet observed.
• On this deployment, access tokens last 15 minutes and refresh tokens rotate with a 30-day lifetime — the connecting tool uses this to renew your access automatically, without asking you to sign in again.
• Your connector consent record (which permissions you approved, and when) is kept for as long as your connection is active. Revoking access does not delete this record. Revocation marks it as revoked, with a timestamp, and Stride retains it as an audit trail of what was granted and when it was withdrawn — the same way a security log keeps a record of a past access grant rather than erasing it.
• A scheduled daily job now exists to purge several other retention-window data classes (evidence, assessment rationale, prepared-post bodies, and similar records), including evidence you attach through this connector. That scheduler's first live run has been directly, independently verified against the dev database (2026-08-10); the pilot database's first run on the same date was reported, not independently re-verified here. So this statement says purging has been observed running on that schedule — verified for dev, reported for pilot — not merely that a scheduler exists.
• Account deletion — not currently an available action. If your underlying Stride account were ever permanently deleted from the database, your connector grants and related records would be deleted along with it, as a database-level guarantee. However, Stride's current account-closure mechanism is administrator-triggered deactivation, not a permanent deletion of the underlying account — deactivation does not trigger that cascade.

Placeholder — do not publish a right-to-erasure claim for this connector until a real, permanent account-deletion path (or an equivalent verified-deletion path) exists and is confirmed — Ryan / Privacy-Legal.

Sign in to Stride and open Manage Connectors to review or revoke this connector's access at any time. Revoking marks the grant as revoked rather than deleting it — see “How long this data is kept” above. You can also revoke access from the connecting AI client's own connector settings (for Claude Code, its MCP server list), independent of Stride.

For a support question about this connector, or to report a security concern affecting it, contact ryan.cappelluti@cgi.com.

If Stride identifies a security incident affecting this connector's data, it is handled under Stride's existing security incident-response process.

Placeholder — the specific notification-timeline commitment to affected end users is not yet decided.

Placeholder — no versioning or change-notice process for this specific statement exists yet. Each published revision from here forward should be dated and a changelog kept.