Observability

Observability

Observability

Metrics, logs, health grades, cost estimates, and alerts for connected Supabase projects.

Observability monitors connected Supabase projects. After a Supabase organization is connected, Dreambase collects each project's metrics, logs, and settings, grades its health, estimates its bill, and reports recommended fixes.

Observability is available to workspace owners and admins. Its pages are in the app sidebar, and every view is also exposed to agents through MCP and REST.

Organizations graph showing connected Supabase organizations and their projects with grades and estimated cost

How Observability works

Connection

Connecting Supabase authorizes Dreambase against one Supabase organization with read access. Every project in the organization is imported with its read replicas and branches, and receives an initial check immediately. Projects created later are imported when the organization is refreshed.

What Dreambase collects

What

How often

Server metrics: CPU, memory, disk, connections, WAL, query load

Every minute, on paid Supabase plans

Logs, busiest routes, statements, Postgres errors, schema changes and deploys

Every 10 minutes per project; logs are read hourly

Health and performance check

Hourly

Full check: security, indexing, errors, capacity, settings and backups

Daily

Minute data is retained for 14 days, hourly rollups for 90 days, and daily rollups for 13 months. All reads come from Dreambase's Observability store; page loads never call Supabase.

Grades

Each project is graded A through F across five weighted categories:

Category

Weight

Security

35%

Performance

20%

Indexing

15%

Errors

15%

Capacity

15%

A critical finding deducts 25 points, a warning 5, and a minor note 1. A fixed finding stops counting against the grade at the next check.

Alerts

Every project is created with a set of preset triggers on critical signals, such as high CPU or disk usage. Presets are off by default and can be turned on and tuned per project. See Triggers.

The pages

  • Organizations: every connected Supabase organization and project, with cost and grade.

  • Projects: one project’s health, recommendations, usage, alerts and settings.

  • Monitoring: live traffic, errors and database load for one project.

Use Observability with agents

Every Observability view is exposed as a tool on the Dreambase MCP server and the REST API under the health-reports scopes: health-reports:read for reads, and health-reports:write to change finding states, run checks, and validate fixes. Agents can also subscribe to alerts through MCP Events. See Official Dreambase MCP.

Observability tools are marked experimental; their arguments and responses can change without notice.

Important boundaries

  • Observability is restricted to owners and admins. Members have no access to its pages or tools.

  • Dreambase has read access to Supabase projects and makes no changes, except fixes explicitly applied by a user.

  • Minute metrics require a paid Supabase plan.

  • Logs are read hourly; the newest log data can lag by up to about 90 minutes.

  • Supabase connections created before Observability may need to be reconnected to grant the required scopes.

Observability monitors connected Supabase projects. After a Supabase organization is connected, Dreambase collects each project's metrics, logs, and settings, grades its health, estimates its bill, and reports recommended fixes.

Observability is available to workspace owners and admins. Its pages are in the app sidebar, and every view is also exposed to agents through MCP and REST.

Organizations graph showing connected Supabase organizations and their projects with grades and estimated cost

How Observability works

Connection

Connecting Supabase authorizes Dreambase against one Supabase organization with read access. Every project in the organization is imported with its read replicas and branches, and receives an initial check immediately. Projects created later are imported when the organization is refreshed.

What Dreambase collects

What

How often

Server metrics: CPU, memory, disk, connections, WAL, query load

Every minute, on paid Supabase plans

Logs, busiest routes, statements, Postgres errors, schema changes and deploys

Every 10 minutes per project; logs are read hourly

Health and performance check

Hourly

Full check: security, indexing, errors, capacity, settings and backups

Daily

Minute data is retained for 14 days, hourly rollups for 90 days, and daily rollups for 13 months. All reads come from Dreambase's Observability store; page loads never call Supabase.

Grades

Each project is graded A through F across five weighted categories:

Category

Weight

Security

35%

Performance

20%

Indexing

15%

Errors

15%

Capacity

15%

A critical finding deducts 25 points, a warning 5, and a minor note 1. A fixed finding stops counting against the grade at the next check.

Alerts

Every project is created with a set of preset triggers on critical signals, such as high CPU or disk usage. Presets are off by default and can be turned on and tuned per project. See Triggers.

The pages

  • Organizations: every connected Supabase organization and project, with cost and grade.

  • Projects: one project’s health, recommendations, usage, alerts and settings.

  • Monitoring: live traffic, errors and database load for one project.

Use Observability with agents

Every Observability view is exposed as a tool on the Dreambase MCP server and the REST API under the health-reports scopes: health-reports:read for reads, and health-reports:write to change finding states, run checks, and validate fixes. Agents can also subscribe to alerts through MCP Events. See Official Dreambase MCP.

Observability tools are marked experimental; their arguments and responses can change without notice.

Important boundaries

  • Observability is restricted to owners and admins. Members have no access to its pages or tools.

  • Dreambase has read access to Supabase projects and makes no changes, except fixes explicitly applied by a user.

  • Minute metrics require a paid Supabase plan.

  • Logs are read hourly; the newest log data can lag by up to about 90 minutes.

  • Supabase connections created before Observability may need to be reconnected to grant the required scopes.

Semantic Skills

Organizations

© 2026 Dream, Inc. All rights reserved.