Project Overseer
Unified portfolio monitoring dashboard showing health across all Wasteology projects. Aggregates pipeline status, digest insights, and task tracking into one view.
Page Overview
Project Overseer
Monitor project health, review pipeline alerts, track 30-day trends, and manage digest tasks
How to Use
Features
Health Metrics
Summary cards showing total projects monitored, active error counts, and overall pipeline status at a glance
Pipeline Alerts
Failed pipeline list with superseded-failure filtering — if the latest run succeeded, prior failures are automatically suppressed
30-Day Trend
Rolling chart of daily health scores to surface slow degradations that single-day snapshots would miss
Digest Tasks
Open and completed task panel with inline close and delete actions for admin users
Health Metrics
The top row of cards provides an at-a-glance portfolio summary pulled from the latest digest snapshot.
| Metric | Source |
|---|---|
| Total Projects | Count of all owned + support projects in the manifest registry |
| Error Count | Unresolved pipeline failures after superseded-failure filtering |
| Pipeline Status | Aggregate pass/fail rate across all owned flow prefixes |
| Digest Age | Time since ops digest daily last ran — warns if stale |
If the digest is older than expected (typically > 26 hours), the page displays a staleness banner. The digest normally runs at midnight via the ADO pipeline scheduler. Investigate if the banner persists through the morning.
Pipeline Alerts
The alerts panel lists every failed pipeline from the latest digest snapshot after superseded-failure filtering has been applied.
Superseded-Failure Filtering
Failures are suppressed when the latest run for the same flow succeeded — meaning the failure was already resolved by a subsequent run. This prevents stale failures from creating noise.
| Alert State | Description |
|---|---|
| ✗Active Failure | Latest run for this flow failed — needs attention |
| ⚠Superseded | Prior failure, latest run succeeded — filtered out by default |
Superseded failures are hidden by default. Toggle the Show Superseded switch to reveal them for debugging purposes.
30-Day Health Trend
A rolling bar/line chart showing daily health scores for the past 30 days. Each day's score is calculated from the digest snapshot captured that night.
Use this chart to:
- Spot week-over-week degradation patterns
- Correlate failures with deployment dates
- Confirm that a remediation actually improved health over multiple days
AI-Generated Insights
The insights panel surfaces the narrative commentary generated by ops digest daily. This is a prose summary of notable events, anomalies, and recommendations produced by the Claude AI model as part of the nightly digest run.
Digest Tasks
The tasks panel shows tasks sourced from the wg_digests.digest_tasks table in PostgreSQL.
Task Actions
| Action | Permission | Effect |
|---|---|---|
| ✓ Close | Admin only | Marks the task as completed in the database |
| 🗑 Delete | Admin only | Permanently removes the task record |
Task mutations (close and delete) require the authenticated user's displayName to be in the DBACCESS_MUTATION_ALLOWLIST environment variable on the server. Non-admin users see the tasks as read-only.
Data Source
All data is sourced from PostgreSQL wg_digests schema, populated by the ops digest daily command running on a nightly schedule via ADO pipelines.
# Run a manual digest to refresh data
uv run ops digest daily