Publish
Git-native publishing
Documentation as a diff. Three formats from one sync, committed to the repo your team already reviews.
How it moves
One fetch, one render, your choice of destination.
AZ Data reads table and column metadata straight from Snowflake, renders it once, and writes the result wherever your team already works — no separate pipeline per target.
Pick one destination or several — the same sync writes to all of them, and nothing is stored on AZ Data's own servers in between.
Markdown
Human-readable pages per schema and table, with business annotations, owners, and the last-synced timestamp at the top.
OKF
A machine-readable knowledge format for tooling that wants structure without parsing prose.
Graph
Tables, columns, and relationships as a graph you can query — the same graph the MCP server answers from.
Your repo, your history
Commits you can review, blame, and revert.
AZ Data pushes to a branch you choose on GitHub or Bitbucket. Every schema change becomes a commit with a readable message, so "who touched this table" is git log, not an investigation.
- Open a PR per change, or commit straight to main — your call
- Docs live next to your dbt models and Airflow DAGs
- Nothing to host: the output is plain files in your repository
Prefer object storage or a wiki instead of Git? The same sync can write Markdown, OKF, and graph files straight to an S3 bucket you own, or publish rendered pages into a Confluence space your team already reads. Git is the default for teams that review docs like code, not a requirement.
Next: serving context to agentsPoint it at a repo. Watch the first commit land.
Onboarding is a Snowflake role and a Git token. Most teams see their first commit within the hour.