The AI & Project Management Study and Research Hub A Joint Initiative of the PMI Brazilian Chapters
Live Platform · MCP Server · Blog · Governance · Site Map & Access Tiers
The AI & PM Research Hub (Núcleo de Estudos e Pesquisa em Inteligência Artificial e Gerenciamento de Projetos) is a multi-chapter research initiative under the PMI® Brazilian ecosystem, advancing the intersection of Artificial Intelligence and Project Management.
Founded in 2024 as a pilot within PMI Goiás, the initiative has grown into a structured alliance of five PMI chapters — PMI-GO, PMI-CE, PMI-DF, PMI-MG, and PMI-RS — with 52 active researchers organized across 7 research streams and 4 strategic knowledge quadrants.
Project Manager: Vitor Maia Rodovalho
| Indicator | Value |
|---|---|
| Active researchers (Cycle 3) | 52 |
| Research streams (Tribos) | 7 |
| PMI chapters | 5 (GO · CE · DF · MG · RS) |
| Events held | 209 |
| Governance entries | 141+ (GC-001 → GC-141+) |
| Blog posts | 9 |
| MCP surfaces | /mcp 342 registry · /semantic 52 intent · /actions 88 overflow |
| Edge Functions | 46 |
| pg_cron jobs | 63 |
| RPCs (SECURITY DEFINER) | 1,066 (1,228 public functions total) |
| i18n keys | 6,700+ (3 locales) |
| Tests | 5,306 passing (5,839 total; DB-aware suite runs with service-role env) |
| Monthly cost | $0 |
| # | Quadrant | Research Streams |
|---|---|---|
| Q1 | The Augmented Practitioner | AI Tools & Ecosystem for PM |
| Q2 | AI Project Management | Autonomous Agents & Hybrid Teams |
| Q3 | Organizational Leadership | TMO & PMO of the Future · Culture & Change · Talent & Upskilling · ROI & Portfolio |
| Q4 | Future & Responsibility | Governance & Trustworthy AI · Inclusion & Human-AI Collaboration |
graph LR
subgraph Client
A[Browser] --> B[Astro 6 SSR]
C[AI Assistant] --> D[MCP Protocol]
end
subgraph "Cloudflare Workers"
B --> E[Pages + API Routes]
D --> F["/mcp · /semantic · /actions"]
E --> G[OAuth 2.1 Server]
F --> G
end
subgraph "Supabase"
G --> H[Auth<br/>Google · LinkedIn · Microsoft]
E --> I[PostgreSQL<br/>1,066 SECDEF · RLS]
F --> J[Edge Functions<br/>46 deployed]
I --> K[pg_cron<br/>63 jobs]
end
subgraph "Observability"
B --> L[PostHog<br/>Analytics]
B --> M[Sentry<br/>Error Tracking]
end
style A fill:#1a1a2e,color:#fff
style C fill:#1a1a2e,color:#fff
style E fill:#F38020,color:#fff
style F fill:#F38020,color:#fff
style I fill:#3FCF8E,color:#fff
style J fill:#3FCF8E,color:#fff
| Layer | Technology | Details |
|---|---|---|
| Frontend | Astro 6 + React 19 + Tailwind 4 | SSR with island architecture, trilingual |
| Hosting | Cloudflare Workers | Edge SSR, OAuth proxy, MCP proxy |
| Database | Supabase PostgreSQL | 1,066 SECURITY DEFINER functions, RLS |
| Auth | Google + LinkedIn + Microsoft | OAuth 2.1, PKCE, native Supabase OAuth server (#1210) |
| MCP | Custom server, 3 surfaces | /mcp (342 registry) · /semantic (52 intent-level, SPEC-280) · /actions (88 overflow for the 256/connector cap) |
| Server Logic | Supabase Edge Functions (46) | Credly sync, attendance, MCP, campaigns, Artia sync, PostHog proxy, AI triage |
| Analytics | PostHog | Product analytics, session replay |
| Errors | Sentry | Real-time error monitoring |
| Cron | pg_cron (63 jobs) | Credly sync, attendance batch, detractor alerts, weekly digests (member · leader · tribe), MCP anomaly detection, LGPD anonymize, log retention, R2 backup, AI retry queues, Artia sync, drive-discover-atas, V4 engagement expiry/anonymize |
| DnD | @dnd-kit | BoardEngine Kanban |
| Rich Text | TipTap | Meeting minutes, blog editor |
Any member can connect Claude, ChatGPT, Perplexity, Cursor, or VS Code to the platform via the Model Context Protocol. All tools are authenticated via OAuth 2.1 (Supabase's native OAuth server, #1210 — the AI client refreshes directly against Supabase Auth on a client-scoped chain, independent of the browser session) with full Row Level Security enforcement.
The server exposes three surfaces (live counts at /health; the tools/list response is the source of truth, never pin a number):
| Surface | URL | Tools | Purpose |
|---|---|---|---|
/mcp |
nucleoia.vitormr.dev/mcp |
342 | Full internal capability registry. Default for clients that accept large catalogs. |
/semantic |
nucleoia.vitormr.dev/mcp/semantic |
52 | Intent-level semantic gateway (SPEC-280 / #1383). One tool per user intent (~7:1 consolidation) with a stable envelope {ok,data,summary,warnings,next_actions,audit}; writes carry authority + confidential-visibility gates as a contract. Use when a strict client rejects the full catalog. |
/actions |
nucleoia.vitormr.dev/mcp/actions |
88 | Overflow surface for the Claude connector's 256-tool-per-connector cap (#1377) — re-exposes the write/action tail the alphabetical cut would drop. Consumed alongside /mcp as a second connector. |
https://nucleoia.vitormr.dev/mcp
sequenceDiagram
participant User as AI Assistant
participant W as Workers Proxy
participant S as Supabase Auth
participant EF as Edge Function
User->>W: POST /mcp (no token)
W-->>User: 401 + WWW-Authenticate
User->>W: POST /oauth/register (DCR)
W-->>User: client_id
User->>W: GET /oauth/authorize
W->>S: Login (Google/LinkedIn/MS)
S-->>W: JWT
W-->>User: authorization code
User->>S: POST /auth/v1/oauth/token (PKCE)
S-->>User: access_token + refresh_token
Note over User,S: Client refreshes directly against GoTrue (#1210)<br/>client-scoped chain; the Worker never refreshes
User->>W: POST /mcp + Bearer token
W->>EF: Proxy with valid JWT
EF->>EF: RLS-enforced query
EF-->>User: Tool result
| Compatibility | Status |
|---|---|
| Claude.ai | Verified |
| Claude Code | Verified |
| ChatGPT | Verified (beta) |
| Perplexity | Verified |
| Cursor / VS Code | Verified |
| Manus AI | Verified (JSON import) |
- Personal workspace with XP, ranking, and Credly badge tracking
- Tribe dashboard with meetings, attendance, and deliverables
- BoardEngine (Kanban, table, calendar, timeline, grouped views)
- Gamification with 10 XP categories
- Trilingual interface (PT-BR · EN-US · ES-LATAM)
- Full board management (create, assign, move, archive)
- Attendance registration and reporting
- Meeting minutes (TipTap rich text)
- Tribe notifications and broadcast
- Admin panel with KPI dashboards and governance
- 28+ Change Requests tracking manual updates
- Stakeholder landing page
- Selection process with blind review
- Sustainability CRUD with financial projections
This project operates under a formal governance model with hierarchical access tiers, a peer review committee (Comitê de Curadoria), and merit-based selection processes. All decisions tracked in the changelog.
- Governance Changelog — 141+ entries (GC-001 → GC-141+)
- Sprint Board
- Contributing Guide
- Zero-Cost, High-Value — All infrastructure on free tiers (Supabase, Cloudflare, PostHog, Sentry)
- Platform as Source of Truth — Member state, gamification, governance, and research outputs live here
- Security by Design — All writes via SECURITY DEFINER RPCs, RLS per member/tribe/role, LGPD compliant
- Data Centralization — Schedules, links, meeting slots in the database — never hardcoded
npm install
npm run build
npm run dev -- --host 0.0.0.0 --port 4321
npm testPrerequisites: Node.js 24+ (nvm), Supabase CLI (≥ 2.100.1 — earlier versions fork-bomb on migration repair, see issue #155 / docs/RUNBOOK.md), Wrangler CLI. See .env.example for variables.
├── src/
│ ├── pages/ # Astro pages (trilingual routes)
│ ├── components/ # React islands + Astro components
│ ├── lib/ # Supabase client, auth, utilities
│ └── middleware/ # CSP, auth, i18n
├── supabase/
│ ├── functions/ # 46 Edge Functions
│ └── migrations/ # Database migrations
├── tests/ # 5,306 passing tests
├── docs/ # Governance, guides, specs
└── scripts/ # Audit and utility scripts
| Document | Purpose |
|---|---|
README.md |
Project entry point (EN) |
README.pt-BR.md |
Versao em Portugues |
README.es.md |
Version en Espanol |
CONTRIBUTING.md |
How to contribute |
AGENTS.md |
Context for AI assistants |
docs/GOVERNANCE_CHANGELOG.md |
All governance decisions |
docs/MCP_SETUP_GUIDE.md |
MCP server setup |
docs/BOARD_ENGINE_SPEC.md |
BoardEngine architecture |
docs/DISASTER_RECOVERY.md |
Backup & recovery |
docs/SITE_MAP.md |
Complete site map with access tiers (trilingual) |
docs/ADMIN_ARCHITECTURE.md |
Admin panel architecture (21 components) |
This repository uses a split license aligned with the Technical Annex governance model:
- Source code is licensed under the MIT License (
SPDX-License-Identifier: MIT). - Documentation and other non-code written materials are licensed under
Creative Commons Attribution-ShareAlike 4.0 International
(
SPDX-License-Identifier: CC-BY-SA-4.0), unless a file or section states otherwise. - Third-party trademarks, logos, credentials, personal data, private operational records, and materials explicitly marked with a different license or access restriction are not covered by these repository-level license grants.
PMI®, PMBOK®, PMP® and PMI-CPMAI™ are registered marks of the Project Management Institute, Inc. This initiative is a collaborative project of independent PMI chapters and is not directly affiliated with or endorsed by PMI Global.