Integrations

Database Access for BI Tools

Generate read-only PostgreSQL credentials, connect Metabase, Tableau, Power BI, or any SQL client to your organization's data, and keep the credentials current as tables change.

Use Database Access when the reports built into DevStride are not enough: connect a BI tool — Metabase, Tableau, Power BI — or any SQL client straight to your organization's data and build your own dashboards.

You generate one set of read-only PostgreSQL credentials for the organization, paste them into the tool, and query.

How it works

Generating credentials creates a dedicated PostgreSQL role for your organization. The role can only SELECT, row-level security policies restrict every query to rows belonging to your organization, and the connection goes to a read replica, so your dashboards never slow down the live application.

Think of it as a window onto the same tables the product uses, with everyone else's data filtered out before you ever see it.

Generate credentials

  1. Open Configure Settings → Integrations → Database Access. The page is headed Manage Database Access. If the entry is missing, your role lacks Manage database credentials.
  2. Select Generate Credentials (the button shows Generating... while it works).
  3. The credential appears as Pending and flips to Active once the role and policies are provisioned. The Connection Details modal opens immediately so you can copy what you need.

Connect your tool

The Connection Details modal offers two ways to copy the connection:

  • Quick Copy — a complete Connection String for tools that accept a URL.
  • Individual Fields — for tools with a form:
FieldValue
Database Typepostgresql
Host, Port (5432), DatabaseAs shown
UsernameThe generated role name
PasswordMasked; reveal it on demand
Schemapublic
SSLEnabled or Disabled, as shown

The modal also suggests a display name for the connection — DevStride - {Your Organization Name} — with a Copy button.

Manage existing credentials

Once a credential is active, the card is titled Read-Only Database Access and shows the role name, a colored Status, the creation date, and these actions:

ActionWhen to use it
View DetailsReopen Connection Details to copy the string or reveal the password. Active credentials only.
Regenerate CredentialsRotate the password. The old connection string stops working immediately, so update every tool that used it. DevStride warns before regenerating.
Sync TablesAppears only when there are tables your credentials cannot read yet — for example after a DevStride release adds one. Until you sync, querying a new table returns a permission error.
Revoke AccessDisable login without deleting anything. Status becomes Deactivated.
Activate AccessRe-enable a revoked credential. Shown only while Deactivated.
DeleteDrop the PostgreSQL role, remove its policies and grants, and delete the record. Cannot be undone.

Status values

StatusMeaning
ActiveReady to use.
PendingBeing provisioned.
SyncingNew tables are being granted.
FailedProvisioning did not complete. Generate again to replace it.
DeletingBeing removed.
DeactivatedLogin disabled by Revoke Access.

Security and scope

  • Read-only. SELECT only; nothing can be changed through these credentials.
  • Your organization only. Row-level security keys every query to your organization.
  • Read replica. Queries never touch the primary database.
  • Gated by one permission. Generate, view, regenerate, sync, revoke, activate, and delete all require Manage database credentials.
  • Statement statistics are private. The pg_stat_statements and pg_stat_statements_info views are not readable with read-only credentials, because they exposed other logins' query counts.

Item type hierarchy

Which item types can sit under which is stored in the work_type_parents table, one row per pairing of an item type and a type it may sit under. An item type can have several parent types (see Item Types), so a type can appear in several rows.

ColumnMeaning
work_type_idThe item type (a row of work_types).
parent_work_type_idA type it may sit directly under (also a row of work_types).
organization_idYour organization.
date_added, date_updatedWhen the pairing was recorded. A type's primary parent is its row with the earliest date_added.

A type with no rows is a top-level type. To list every type with its parents:

select child.name as item_type, parent.name as parent_type
from work_type_parents p
join work_types child on child.id = p.work_type_id
join work_types parent on parent.id = p.parent_work_type_id
order by child.name, p.date_added;

Status table renames

DevStride has renamed its status vocabulary in the read-only database, as announced in the 2026-09-07 notice:

Old name (still works)New name
lanesstatuses
lane_collectionsstatus_collections
lane_id (on workitems, item_transactions)status_id
lane_rank (on workitems)status_rank
lane_collection_id (on boards, teams, board_folders)status_collection_id
automate_lanes (on organizations)automate_statuses
service_desk_lane_settings, service_desk_reopen_lane_id (on work_types)service_desk_status_settings, service_desk_reopen_status_id
type lane_typestype status_types

The old table names are now read-only views with identical columns, and every old column is kept equal to its new twin, so dashboards and scheduled queries keep returning the same results. Your credentials can read both names; if a new name returns a permission error, select Sync Tables.

Workstream tables

Old name (still works)New name
foldersworkstreams
folder_count, archived_folder_count (on folders)workstream_count, archived_workstream_count (on workstreams)
root_folder_number (on workitems, notifications)root_workstream_number
next_folder_number (on item_counters)next_workstream_number

folders is now a read-only view over workstreams with identical columns, including folder_count and archived_folder_count, and each old *_folder_number column is kept equal to its new twin, so dashboards and scheduled queries keep returning the same results. Your credentials can read both names; if workstreams returns a permission error, select Sync Tables.

History and notification values

WhereOld value (no longer stored)New value
item_transactions.transaction_typeadded_to_lane, removed_from_laneadded_to_status, removed_from_status
notifications.actionlane_modifiedstatus_modified
notifications.summary (item notifications)Lane modified, New folder created, Folder description modifiedStatus modified, New workstream created, Workstream description modified
activity_logs.datalaneId, laneRank (the field of a change, and the keys of the before and after snapshots); rootFolderNumber (in itemMetaData)statusId, statusRank; rootWorkstreamNumber

Older records were rewritten once, in place, so one new value now covers the whole history: you do not need to match both spellings.

Workstream values in history, jobs, automation runs and imports

WhereOld value (no longer stored)New value
activity_logs.typefolderworkstream
activity_logs.dataitemType and a relationship's entity of folderworkstream
jobs.input, jobs.output, jobs.progressitemType of folder; in recalculation jobs, the folders count, the folder entity type and the folder_count, archived_folder_count field namesworkstream; workstreams, workstream, workstream_count, archived_workstream_count
workflow_runs.inputitemType of folderworkstream
import_job_runs.import_result, import_job_runs.errorsitemType of folderworkstream

Only these system-written values changed: text people typed (descriptions, messages, custom field values) and the input a webhook sent to an automation are stored as they were. Older records were rewritten once, in place, during the 2026-11-16 release; a small number of records written while that release was rolling out may still read folder until a later clean-up rewrites them too.

Removal of the old names

The legacy status and workstream names are removed no earlier than 90 days after the 2026-09-07 notice sent to every organization with active credentials — so not before 2026-12-06 — and only after one further release; you will receive a second notice before removal. If you maintain dashboards or scheduled queries against these tables, move them to the new names during that window.

  • GitHub
  • Connect AI & the MCP Server — the MCP tools get_database_credentials, generate_database_credentials, and delete_database_credentials manage the same credential