Maple DOGE

A proof-of-concept OSINT app for Canadian public spending, procurement, charity, registry, governance, policy, and adverse-media data.

Maple DOGE is built to support human reviewers, not replace them. The app ranks and explains public-spending signals, then routes cases into review queues, case workspaces, advisory actions, briefs, and outcome notes so a person can verify sources before any policy or operational decision is made.

Maple DOGE turns fragmented public-sector datasets into evidence-backed investigation modules.

The app connects entity search, dossiers, ranked watchlists, procurement analytics, policy-gap review, graph exploration, adverse-media context, and a human-in-the-loop review workflow. The goal is not to make automatic enforcement decisions. The goal is to help a reviewer understand which public-spending cases deserve attention first and why.

flowchart LR
  UI["React + Vite frontend"]
  VM["Azure VM web server"]
  API["Node/Express API"]
  PG["Postgres source/serving data"]
  BQ["GCP BigQuery analytics"]
  EXT["Official public datasets"]

  EXT --> BQ
  EXT --> PG
  UI --> VM
  VM --> API
  API --> PG
  API --> BQ

Built the strongest full workflow in the PoC.

• Matched funding recipients to registry status using business number roots.
• Separated registry-backed cases from lower-confidence funding-only fallback cases.
• Added score bands, caveats, source context, advisory actions, action briefs, and outcome tracking.

• Flagged recipients with weak capacity signals.
• Used missing BN, sparse grants, high average award value, and multi-department patterns.
• Added recipient detail pages for evidence review.

• Detected circular giving/funding paths using CRA relationship data.
• Ranked loops by hop count, participant count, bottleneck, and flow value.
• Rendered loop paths as graph evidence.

• Compared original federal agreement values to latest cumulative amended values.
• Avoided raw summing of federal amendment rows because agreement values are cumulative.
• Linked Alberta competitive contracts to sole-source follow-ons by normalized vendor name.
• Flagged high growth, follow-on value, near-threshold patterns, repeated relationships, and nonstandard justification context.

• Built BigQuery concentration outputs by source, department, and category.
• Calculated HHI, CR4, top supplier share, effective competitors, total dollars, and top entities.
• Added invariant checks so impossible share metrics cannot be published.

• Normalized director/person names.
• Built organization-pair and person-detail views.
• Used graph layouts to show shared governance relationships.

• Used official planning, mandate, performance, infrastructure, housing, and health indicator sources.
• Compared stated priorities and planned spending against observed amounts, results, or delivery status.
• Added confidence levels and caveats to prevent overclaiming.

• Split the work into two streams.
• 8A finds overlapping funding across federal grants, Alberta grants, and CRA-reported government funding.
• 8B reviews priority gaps, project delays, spending variance, and allocation-without-project-match cases.
• Added public-sector caveats for expected co-funding.

• Built procurement-grade views from contract datasets.
• Decomposed growth into contract count, average contract value, amendment contribution, interaction effects, and vendor concentration.
• Clearly labels the metric as average contract value, not unit price.

• Moved media scanning to the backend.
• Used Google News RSS and NewsAPI.
• Removed browser-side public CORS proxy dependency.
• Added graceful failure behavior so failed scans do not appear as clean results.
• Treated media as contextual review input, not a standalone decision trigger.

Maple DOGE is advisory by design.

src/
  React frontend, routes, dossiers, challenge pages, graph components

backend/general/
  Node API, data access, entity-resolution scripts, challenge ingestion scripts

backend/general/visualizations/server.js
  Main JSON API used by the web app

deploy/
  Azure VM deployment helpers

public/
  Static assets, including the Maple DOGE logo

Maple DOGE demonstrates that a public-spending accountability app can combine:

• public data ingestion
• entity resolution
• ranked review signals
• source-linked dossiers
• graph evidence
• procurement analytics
• policy analytics
• GCP BigQuery analytical tables
• Azure VM deployment
• human-in-the-loop review workflows

Challenge 1 is the most complete decision workflow. The other challenges are implemented as investigation modules with ranked outputs, graphs, tables, charts, source caveats, and analytical evidence ready for deeper validation.

Copyright (c) From 2026-3000 dirm02. All rights reserved.

This project is protected by the repository COPYRIGHT notice. No permission is granted to use, copy, modify, publish, distribute, sublicense, or sell any part of the software or source code except as required by applicable law or with prior written permission from the copyright holder.