Access Code Timeline in the Seam Console

Access Code Timeline in the Seam Console

Built for debugging

Coming to more resources

2026.05.04 Changelog

New

• Spaces ↔ connected accounts: New endpoints POST /spaces/add_connected_account and POST/DELETE /spaces/remove_connected_account let you scope a connected account to a specific space, and /connected_accounts/list accepts a space_id filter.
• customer_data on spaces: The Space resource now exposes customer_data with time_zone, default_checkin_time, and default_checkout_time. /spaces/create accepts the field, and /spaces/update patches keys individually (pass null to clear).
• use_backup_access_code_pool on /access_codes/create_multiple: The bulk-create endpoint now honors the backup access code pool with the same constraints as /access_codes/create (time-bound only, supported providers only).
• requested_at on response errors: Error responses now include a requested_at timestamp, making it easier to correlate Seam errors with your logs and provider-side activity.
• Inspect request error details: Every error response now includes a request_error_id, and the corresponding error-detail page lives at a public URL (/public/request_error/<id>) so you can open it directly — and share it with Seam support — for richer context on a failed request.
• Pagination on /spaces/get_related: The endpoint now paginates results per resource type (capped at 50 per type), giving you cursor-based control over related-resource lookups and resolving query failures on workspaces with very large access-method counts.

• Redesigned billing page and upgrade flow: The Console billing page has been redesigned, and the upgrade flow refreshed.

• Dormakaba Ambiance mobile keys: Mobile key support is now wired up for Dormakaba Ambiance, including a new can_issue_mobile_keys: true capability flag and end-to-end issuance for mobile-key-only access grants.
• Dormakaba Oracode — native time-bound codes: Oracode codes are now classified as native time-bound and program on creation. Codes scheduled up to a year in advance now report creating (not deferring_creation) and register with Dormakaba on the next sync, instead of waiting for a 1-hour pre-start window.
• Brivo — PIN and mobile pass credentials: Credential discovery now includes PIN and mobile pass credentials in addition to cards. PINs map to access_method: "code", mobile passes to cloud_key, and cards to card.
• Salto KS — salto_ks_lock_access_code_support_removed warning: When a Salto KS lock's device model changes from one that supports access codes to one that doesn't, the device now carries a clear warning with a link to the help-center article. The warning auto-clears if the lock model is restored.

• DeviceDB now indexed in the public Seam MCP Docs Server: The hosted MCP at mcp.seam.co/mcp now indexes 400+ device models and full manufacturer / provider docs, plus per-device-type catalog chunks (sensors, smart locks, thermostats, etc.) — so questions like "which noise sensors does Seam support?" return the full inventory.

Changed

• deleting pending mutation visible while a delete is in flight: /access_codes/delete now writes a deleting entry to pending_mutations_map atomically with is_being_deleted = true, so /access_codes/get and the timeline reflect the in-flight delete until perm-delete completes.
• Reason-aware error messages on access-code failures: When an access code fails because the device is offline, the connected account is disconnected, or the device has been removed, the error_map message on failed_to_set_on_device / failed_to_remove_from_device now reflects the real cause instead of generic boilerplate.
• max_access_codes_reached warning fires from provider responses: Provider-reported "all PIN slots are full" errors now correctly raise the max_access_codes_reached warning on the device — previously the warning only fired from Seam's count-based pre-check.

• Salto KS — salto_ks_user_not_subscribed warning surfaces on subscription-limit errors: When Salto returns the 1101 subscription-limit error, the affected access code now carries salto_ks_user_not_subscribed instead of being silently cleared.
• ASSA ABLOY Visionline — provider error details preserved on card encoding 400s: The provider's code, message, and developerMessage are now surfaced in the credential's error_map instead of being collapsed into a generic error.

Fixed

• Pre-create updates clear pending mutations: Updating an access code before it's been programmed to the device now correctly clears updating_code / updating_name / updating_time_frame once the create succeeds.
• Native time-bound codes outside the grace window report deferring_creation: Far-future native time-bound codes (Schlage, Yale, August, Dormakaba) now report deferring_creation until inside the 72-hour grace window, instead of misleadingly reporting creating.

• Smarter disconnect handling: Brivo accounts no longer flip to disconnected on transient rate-limit responses, and the acs_system_disconnected warning clears once the account reconnects. Genuine refresh-token failures correctly disconnect the account and emit connected_account.reauthorization_requested so your app can prompt a reconnect. Note: we sent an advisory to all Brivo customers last week to reconnect their Brivo accounts so these changes take effect.

• Backup access code pool no longer accepted: Oracode PINs are derived algorithmically by the door firmware from the stay, so backup-pool semantics don't apply. /access_codes/create with use_backup_access_code_pool: true on Oracode now returns a 400, and /devices/get reports properties.supports_backup_access_code_pool: false.

• subscription_limit_almost_reached warning now clears correctly when the user count drops back below 80% utilization.
• Self-heal access codes mid-create: Codes whose subscribe step failed after the user was created now self-heal on the next sync instead of staying stuck.
• Robust handling of access-group changes: ACS sync no longer crashes when an access group is deleted between the list and the read, and the access-group lifecycle is more resilient when the upstream API fails partially.
• Empty-sites accounts no longer hit Salto in a tight loop: Connected accounts where all sites have been removed no longer make repeating 401 calls to Salto.

• Recover already-created codes on duplicate-code errors: When Schlage reports AccessCodeAlreadyInUse, Seam matches the existing code on the lock by PIN and binds the local row to it, eliminating a large class of duplicate_code_on_device errors.
• Skip push when the device or account is disconnected: Push attempts now short-circuit on both device-level and account-level disconnection, so Seam stops calling Schlage with stale tokens for orphaned codes.
• Native time-bound codes respect IANA timezones: Timezone strings (America/New_York, etc.) reported via /devices/report_provider_metadata are now resolved correctly, fixing native time-bound codes that were being shifted by the UTC offset on locks with location.timezone set.

• Drop invalid joiner cards before encoding: Mobile-key requests now strip joiner cards Visionline can't accept (overridden, cancelled, discarded, expired, unknown) before sending — eliminating a class of stuck mobile keys with illegal_card_in_joiners errors.
• Card and mobile key starts align for dual-access reservations: Dual-access reservations (card + mobile key) no longer fail when the front desk encodes the card after starts_at; both credentials now align on a starts-no-earlier-than-now timestamp.

• No more false external-deletion on Schlage-via-SmartThings: Schlage locks behind SmartThings sometimes return an empty lockCodes attribute right after a setCode / deleteCode. Seam no longer treats that signal as an external deletion.

• Webhooks with non-UUID partnerUserID ignored: August webhooks carrying a non-UUID partnerUserID no longer enter the matching flow.
• PIN operations track end-to-end so webhooks resolve correctly: Pre-confirming deletes was masking real failures; deletes now wait for the actual partnerUserID, and pin operations are tracked through their lifecycle so webhooks reliably resolve to the correct access code.
• Managed access codes no longer adopt external partner PINs: A legacy August path could have a managed access code adopt an external partner's PIN, leaving the managed code stuck unset. Adoption is now blocked when the existing August user belongs to a different partner. Yale Home /pins deletion is similarly gated on the partner-user link existing.

• No disconnect on proactive token refresh failure: A failure during proactive token refresh no longer flips the account to disconnected.
• Device ownership preserved for shared Nuki Web locks: Devices shared via Nuki Web no longer get reassigned away from their owning account during sync.

• No false low-battery events on mains-powered devices: AC-powered Akiles devices no longer emit false device.low_battery events. Battery level is now only computed when the device is actually battery-powered.

• Manual thumbturn lock/unlock events now surface: Manual thumbturn events were being dropped on negative-UTC devices and by case-sensitive event mapping. Both are fixed; manual thumbturn now reliably emits lock.locked / lock.unlocked.

• No more online/offline flap from competing pollers: A second poller was overwriting the authoritative connection state, producing spurious device.disconnected → device.connected event pairs every ~10 minutes. The device-level poll no longer writes online state.