slashpeak
Product

Audit logs

One timeline of SlashPeak scans, BigQuery Cloud Audit Logs, and every Cognis AI and AI DBA action. On Growth, Pro and Enterprise.

The Audit Logs page (/audit-logs) is one timeline of who did what, and when. On the trial and Starter it shows a lock.

PlanAudit Logs page
30-day trial, StarterNo
Growth, Pro, EnterpriseYes

What is on the timeline

Pick a source at the top, or All:

SourceEvents
SlashPeakScans for this connection: started, finished, failed, and scan data flushes
BigQueryCloud Audit Logs for BigQuery: Admin Activity (who changed datasets, tables, IAM) and Data Access entries. Needs extra setup, below
CognisEvery AI query optimizer call, every AI DBA model call, every AI DBA step (plan, policy decision, approval, BigQuery call, pull request, roll back), Cognis Settings changes (policy, operator key, pause, AI consent), and GitHub and dbt Cloud events

Cognis events are written to an append-only log. A row cannot be edited. Secrets are removed before a row is written, and the log never holds chat message text or a full AI prompt. Each Cognis event is kept for 12 months, then deleted.

Filters

Filter by source, project, category, outcome (started, success, failure, denied) and actor (a person, the policy, the worker, GitHub, or the system). Save a filter set as a segment to reuse it.

Cognis → Activity shows the same Cognis events on their own.

Add BigQuery Cloud Audit Logs (optional)

Core scans do not need this. Turn it on when you want Google's own record of changes next to your job history.

  1. Enable the Cloud Logging API. See Enable Google APIs.
  2. Grant logging.privateLogEntries.list in the optional custom role. See IAM permissions.
  3. Make sure Data Access audit logs are turned on for BigQuery in the project. IAM alone is not enough: if Google does not write the logs, the feed stays empty.

When it works, BigQuery rows carry the badge From Cloud Audit Logs.

What it is not

  • It is not a SIEM and not a log archive. It is context for expensive or surprising jobs and for AI actions.
  • We do not become your log sink, and we do not keep GCP logs longer than the product needs to show this page.
  • Job metadata from INFORMATION_SCHEMA stays the main source for usage and cost.

On this page