Schema analysis

Experimental analysis surface

Analysis output is outside the stable migration contract. It provides findings for review; it is not an automatic schema repair or a replacement for versioned migrations and deployment checks.

The installed CLI command is aksara ai schema-health. Run it from an application with its model registry and PostgreSQL connection configured:

aksara ai schema-health --format json
aksara ai schema-issues --format json
aksara ai schema-issues --severity danger

Health statuses are healthy, degraded and danger. Issue severity uses info, warning and danger. Filters for schema-issues include --severity, --kind, --table and --app-label; inspect --help for the exact options.

The former aksara ai doctor and aksara ai doctor --fix examples were not valid v0.7.0 commands. There is no SchemaDoctor class to instantiate. Do not interpret a generated suggestion as an applied migration.

Python integration

The actual async API is:

from aksara.ai import analyze_schema_health

report = await analyze_schema_health(app)
print(report.status)

This is an integration snippet, not a standalone program: app is your running application, the database must already be connected and the intended models must be registered. It compares the model schema map with database introspection. A missing database is a configuration finding, not proof that all application tables are absent.

Review the concrete findings against your migration history. Generate and review a migration using the migration guide, then apply it through the production deployment process. Schema analysis cannot establish authorization, correct RLS policy, backup recoverability or the safety of an arbitrary migration.

For production security posture use Doctor. For durable schema and registration readiness use the separate durable preflight.