GitReport Pipeline — Value Delivered

What We Built

Multi-Source Data Ingestion Pipeline

A 13-step extraction pipeline pulling from 6 data sources, normalizing into a unified entity hierarchy, and auto-generating 8 executive report routes every 30 minutes.

6 Data Sources

GitHub API — ~95 repos, issues, PRs, comments
project-tracker — projects, sub-projects, notes
Stripe API — revenue, subscriptions, payments
CSA Scanner — 30 engagement artifacts
Hook Queue — Claude Code session logs
Asana MCP — task queue with HITL approval

Automated Schedule

Every 30 min — Flywheel: sessions + issues + reports
2:30 AM daily — Stripe revenue extraction
Sunday 7pm — Weekly executive handoff
On demand — Full 13-step bootstrap
Per session — Asana HITL review

Data Flow: Sources → Vault → Reports

Report System

8-Route Executive Report System

Each route serves a different operational persona with metrics computed from the same vault data. All regenerate automatically every 30 minutes during work hours.

#	Route	Persona	Key Metrics
1	Client Outcome	Revenue Ops Analyst	Deal velocity, payment health, churn risk
2	Activity Pipeline	DevOps Lead	DORA metrics, completion-to-delivery ratio
3	Client Pipeline	Customer Success Manager	Client health A-F, delivery cadence
4	Pipeline Outcome	Project Manager	Progress vectors (biz/user/market/tech/ops)
5	Task Dashboard	Operations Director	Unified priority queue, cross-route scoring
6	Financial Value	Financial Analyst	$ allocation, VP scoring, activity state
7	Value Dashboard	Operations Director	Risk signals, PROTECT/CONVERT matrix
8	Weekly Handoff	Operations Director	Cross-route synthesis, action labels

Report Route Architecture

Value Attribution

Value Attribution Engine

A weighted scoring system connecting every code activity to business value. Activities earn Velocity Points (VP) based on type and completion state, then revenue is allocated proportionally.

Activity Multipliers

Feature — 3.0x VP
Bug Fix — 2.0x VP
CSA Converted — 2.0x VP
Demand Gen — 1.5x VP
Chore — 1.0x VP
CSA Unconverted — 0.5x VP
Issue Opened — 0.25x VP

Risk Classifications

Protected Revenue + VP > 0
Unattributed VP > 50, $0 Stripe
Dormant VP = 0, $0 Stripe

Action Labels

PROTECT — Revenue at risk
CONVERT — Velocity without revenue
ACCELERATE — Momentum strong
UNLOCK — Needs engagement

Activity Type Distribution (253 Total)

Production Metrics

What We Delivered — By The Numbers

11Days to Production

7,772Lines of Code

651Vault Entities

264Issues Extracted

90+Session Notes

30CSA Engagements

$5K/moRevenue Tracked

699Velocity Points

Before → After

Before GitReport	After GitReport
No visibility into cross-client activity	Real-time dashboard across 14 clients, 95 repos
Manual revenue tracking	Stripe-verified per-activity attribution
No risk detection	Automated PROTECT/CONVERT/ACCELERATE signals
Session work undocumented	90+ session notes auto-ingested with client attribution
Weekly status unknown	Auto-generated executive handoff every Sunday

Client Portfolio

Client Value Matrix — March 2026

Client	Monthly $	VP (30d)	Activities	Status
HHE	$5,000	100	28	Protected
Kaleido Life	—	101	26	Unattributed
FLY	—	100	39	Unattributed
AAS	—	100	5	Protected
Melody Keel	—	100	0	Unattributed
JuxtMedia	—	100	21	Protected
SecondAct	—	98	120	Unattributed

Client Portfolio — Revenue x Velocity

Client Input Needed

Strategic Decisions — Next Phase

Four independent decisions that shape GitReport's evolution. No ordering dependency — all can be resolved in parallel.

1. Pipeline Expansion Priority — Which data source next?

Recommended: A — Client analytics / product usage metrics
Demand-side signals are the #1 gap. Supply-side data is mature — we know what was built, but not whether it was used or validated.

A. Client analytics (demand-side signals)
B. Additional GitHub orgs (client repos)
C. Financial forecasting (budget data)
D. Deployment monitoring (Railway/Vercel)

2. Report Distribution — How should weekly handoffs reach stakeholders?

Recommended: A — Email digest
Lowest friction. Weekly cadence already exists. Obsidian inbox is agent-only — email extends reach to human stakeholders.

A. Email digest to stakeholders
B. Slack channel post (via slack-relay)
C. Published HTML artifact (via CSA pipeline)
D. Keep current (Obsidian inbox only)

3. Vault Access Model — Should GitReport be queryable by other systems?

Recommended: A — Read-only API endpoint
Enables agent-to-agent queries. FLY sales bot could pull client context from GitReport. Localhost only — no external exposure.

A. Read-only API (localhost, agent consumption)
B. Obsidian Publish (team visibility)
C. Keep current (agent + file access only)

4. Automation Depth — Should Asana HITL auto-approve known mappings?

Recommended: A — Auto-approve when confidence > 90%
asana-client-map.json already learns patterns. High-confidence mappings don't need manual review — reduces per-session friction.

A. Auto-approve (>90% confidence)
B. Always manual
C. Hybrid (P0/P1 auto, others manual)

What's Next

Deferred Priorities & Dependencies

Decision 1
Analytics Source

→

DP-1
Demand-Side Signals

→

DP-2
Shipped vs Validated

Item	Status	Activation Gate
Demand-side signals integration	Blocked on Decision 1	Analytics source identified + access granted
Shipped-vs-validated distinction	Blocked on DP-1	Demand data flowing for >7 days
Additional GitHub users/orgs	Blocked on client access	Client grants org membership or token
Redis queue replacement	Not yet needed	SQLite lock errors observed in flywheel logs