DataGuard AI logoDataGuard AIAutonomous data stack protector
All quiet — nothing is breaking right now

Hey there 👋 your data stack is protected.

I watch every pull request, trace the blast radius through DataHub lineage, and open a reviewable fix PR before a rename ever reaches production. Here's what I've been up to.

Simulate a breaking changeOpen lineage repair

17

Schema Breaks Prevented

+4 this week

86

Downstream Lineage Impacted

entities monitored

12

Auto-PRs Merged

code fixes generated

100%

DataHub Context Sync

Healthy · fully synced

Recent activity

Every action includes the reason I took it.

Live

  1. GitHub PR #42 renamed column user_id to account_id

    2 minutes ago

    DataGuard AI tagged 14 downstream dbt models in DataHub and opened fix PR #43 with verified lineage context.

  2. Breaking change detected in stg_customers

    18 minutes ago

    billing_address dropped in favour of primary_address — 5 downstream assets flagged before merge.

  3. Glossary term canonicalized

    1 hour ago

    primary_address published to the DataHub business glossary with an owner-approved definition.

  4. DataHub context sync completed

    3 hours ago

    1,284 entities re-indexed · lineage freshness 100% · 0 drift warnings.

  5. Airflow DAG customer_360_daily auto-patched

    Yesterday

    Column reference list updated and re-validated against the DataHub schema before the 06:00 SLA window.

Agent crew

Three specialists, one MCP connection to DataHub.

  • Impact Agent

    Traverses DataHub lineage to score what a diff will break.

  • Healing Agent

    Rewrites dbt / Airflow references and drafts the fix PR.

  • Governance Agent

    Writes tags and glossary terms back into DataHub.

MCP tools in use

  • get_lineageTrace downstream blast radius
  • add_tagsMark assets 🏷️ under-migration
  • update_glossary_termCanonicalize column meaning

Auto-fix pull requests

polling

Tell me which repository to watch (e.g. acme/analytics) and I'll track every pull request I open there.