Production DevOps data warehouse: 6 live sources, 619 entities, 7 intelligence reports with auto-discovery extraction, 30-minute automated refresh. Built in 8 days, validated against live handoff data.
In 8 days, we built GitReport — a production DevOps data warehouse that ingests data from 6 live sources, models 619 entities across a 6-level hierarchy, and generates 7 domain-specific intelligence reports every 30 minutes. Auto-discovery extraction means new repos with issues are automatically detected — no manual configuration needed.
Every data source that matters for operations intelligence now flows into a single unified queue.
Full metadata, issues, and PR history
REST API extraction with subproject depth
Live revenue, subscriptions, payment history
Call-Specification-Artifact engagement inventory
Agent-issue-hook DevOps pipeline activity
Hook queue ingestion of agent activity
The vault isn't a flat file dump — it's a 6-level hierarchy with wiki-links creating a navigable knowledge graph. Auto-discovery extraction detects new repos automatically — every entity is Obsidian-compatible with YAML frontmatter and bidirectional links.
Each report speaks in the language of a specific operations role, surfacing the metrics that role cares about most. Auto-generated from vault data every 30 minutes.
| Route | Persona | Key Question | Mechanism |
|---|---|---|---|
| Client Outcome | Revenue Ops Analyst | Which clients are paying? At risk? | Revenue outcome scoring |
| Activity Pipeline | DevOps Lead | How fast are we shipping? | Agent-adapted DORA metrics |
| Client Pipeline | Customer Success Mgr | What’s each client’s health? | Weighted health scoring (A–F) |
| Pipeline Outcome | Project Manager | Progress across 5 vectors? | business / user / market / tech / ops |
| Financial Value | Financial Analyst | Where is every dollar going? | $ allocation, VP scoring |
| Value Dashboard | Operations Director | What needs protection vs. conversion? | PROTECT / CONVERT / ACCELERATE |
| Weekly Handoff | Executive Ops | What happened this week? | Cross-route synthesis + narrative |
The system identifies where money is at risk and where effort lacks revenue backing. Four action categories drive prioritized next steps.
The vault stays current without human intervention. A cron job runs 7 extraction and generation steps every 30 minutes from 8am–10pm.
Shell + Python stdlib + SQLite. No frameworks, containers, or cloud services.
URL-exact (46), name-fuzzy (3), path/git-remote. 49 cross-source matched pairs.
New repos with issues are detected automatically via GitHub API. No manual REPOS list to maintain.
W11 handoff cross-validated against vault data. Found and fixed 3 systemic visibility gaps in one session.
Single-command orchestration from data extraction through report generation. 12 pipeline steps, 16 scripts, one vault.
| Item | Priority | Status | Dependencies |
|---|---|---|---|
| Asana HITL integration | P1 | Spec’d | Asana MCP tools |
| Weekly handoff cron (Sun 7pm) | P1 | Ready | Cron scheduling |
| Prospect-aware scoring | P1 | Gap | Melody Keel misclassified as Dormant |
| Demand-side signal layer | P2 | Gap | Product analytics source |
| Client-facing value snapshots | P2 | Ready | CSA pipeline |
| Session-to-client attribution | P2 | Gap | 0 sessions tagged to any client |
Want to weigh in on priorities?
→ Open the Input Form to provide your decisions on reporting cadence, demand signals, client-facing artifacts, and Asana integration depth.