Onboard a new dev in a day
A junior reads the documentation, sees the tables, the columns, the descriptions, and the recent changes. No tribal knowledge required.
Pillar 01 · Documentation
Documentation that follows your schema.
Taavik scans the database and keeps an internal Markdown documentation layer aligned with the latest structure. New columns appear in the doc. Renamed columns update in place. Manual notes you wrote stay where you put them.
What you get
A real data dictionary, kept honest by every scan.
Most database wikis are wrong within a week. Taavik fixes that by making the documentation a function of the schema. The agent scans, the diff updates the doc, the change history records when it happened.
Generated
From the schema, after the first scan
Tables, columns, types, nullability, defaults, indexes, foreign keys. All the technical metadata, formatted as readable Markdown.
Updated
Automatically when the schema changes
A scan detects a new column. The documentation includes it. A column gets renamed. The doc renames it in place. No manual sync.
Yours
Manual notes preserved
Add a description to a table, an explanation to a column, a known-issue paragraph. Auto updates touch the technical metadata only and never overwrite your prose.
Audited
Change history per object
The doc shows when a column was added or modified. New hires read the latest state and the recent history in one place.
Sample
What the generated Markdown looks like.
Plain Markdown rendered inside the workspace. Predictable structure so search and links keep working across updates.
# Database: production ## Schema: public ### Table: customers Customer master data used by the application. | Column | Type | Nullable | Default | Description | |---|---|---:|---|---| | id | uuid | no | gen_random_uuid() | Primary identifier | | email | varchar(255) | no | | Customer email address | | created_at | timestamp | no | now() | Creation timestamp | #### Change history - 2026-05-09: table detected for the first time - 2026-05-12: column `email` changed from varchar(128) to varchar(255)
Why teams install it
Three jobs the wiki used to do, and stopped.
Settle the meaning of a column
When two teams disagree on what `status_code` means, write the description once. It survives the next migration.
Audit what changed and when
Compliance asks for a data dictionary with history. Generate it without rebuilding it from a backup.
Documentation guides
Start with the pages your team already needs.
PostgreSQL schema documentation
Generate current object pages from PostgreSQL metadata and keep them aligned after migrations.
Read the guideSQL Server database documentation
Document SQL Server tables, columns, keys, views, and recent changes from a private scan.
Read the guideLiving database documentation
Use generated facts and preserved notes to avoid stale database docs.
Read the guideKeep docs up to date
A compact workflow for scan cadence, change review, and documentation ownership.
Read the guideFAQ
Living database documentation
Will my manual notes be overwritten when the schema changes? +
No. The auto-update is scoped to the technical metadata: column types, nullability, defaults, foreign keys. Your prose paragraphs stay untouched, and a manual note is preserved through schema renames where the underlying object identity is the same.
Where does the documentation live? +
Inside the workspace. The Markdown is stored in the cloud control plane and rendered in the browser. There is no external integration in v1, no automatic export. The point is one reliable place, not yet another sync target.
Can the documentation be exported? +
v1 keeps the documentation internal on purpose. Exports are on the roadmap. If you need a one-off Markdown snapshot for an audit, contact us.
Does it support multiple databases? +
Yes. One workspace, many connections, many databases. Each database has its own documentation tree and its own change history.
How fresh is the documentation I am reading? +
The header of each page shows the snapshot id and the last-scan timestamp. You always know how recent the data is. Pro plans scan as often as every hour.
Try it
Open the workspace and watch the documentation generate itself.
Free for one connection, forever. The first documentation lands within ten minutes of the first scan.