Skip to content
← All projects

Full Stack

WorkMesh

Permission-aware company visibility

Status
Complete
Year
2026
Live application
Runs locally with Docker Compose. Not publicly deployed.
Company Map showing people and teams as a graph
p50, employee dashboard (300 requests, concurrency 10)
80 ms
p50, employee dashboard (300 requests, concurrency 10)
errors across the latency benchmark
0
errors across the latency benchmark
tenant tables under row-level security
35
tenant tables under row-level security

01 Problem

In a small company, managers often don't know what everyone is working on and employees know the work of only a handful of colleagues. WorkMesh connects company → departments → teams → people with goals → initiatives → projects → milestones → work items, with permission-aware visibility and no surveillance.

02 Product overview

A modular NestJS monolith on PostgreSQL with row-level security powers an interactive Company Map, five role-based workspaces, roadmaps, a blocker state machine, a dependency graph with impact analysis and deterministic meeting briefs. Notifications flow through a transactional outbox, BullMQ and Socket.IO, re-authorizing every recipient before delivery.

Implemented

  • Interactive Company Map (React Flow + ELK) with People and Work modes
  • Five role workspaces: Employee, Manager, Project lead, Leadership, Admin
  • Blocker state machine, roadmaps (timeline, board, list) and a dependency DAG with cycle prevention
  • Deterministic meeting briefs with Markdown and print export
  • Full-text search with authorization applied in SQL
  • Admin permission inspector

Planned, not yet built

  • Read-only calendar integration
  • Per-field visibility policies
  • SSO / SCIM
  • Slack and Teams digests
  • Historical health trends

03 System architecture

RESTSET app.org_idNext.jscookie + CSRFGuardsauth · CSRF · tenant txPolicy engine@workmesh/corePostgreSQLFORCE RLSOutbox → BullMQre-authorizeSocket.IOIDs only
Every request runs inside a tenant transaction; row-level security backs the policy engine as defense in depth.clientservicestoremodelqueue

04 Core technologies

Backend

  • NestJS 11
  • Prisma 6
  • PostgreSQL 16 (RLS, FTS)
  • Redis 7
  • BullMQ
  • Socket.IO
  • Zod
  • Argon2id

Frontend

  • Next.js 16
  • React 19
  • React Flow + ELK.js
  • TanStack Query & Table
  • Zustand
  • Recharts

Quality

  • Vitest
  • Jest + Supertest
  • Playwright
  • gitleaks
  • Docker

05 Engineering challenges

Authorization everywhere

Visibility rules apply to search, graphs, notifications and dashboards alike, so one pure policy engine is shared across the API and workers.

Realtime without leaks

Realtime events carry only IDs; clients fetch data through the same authorized API, and each recipient is re-authorized before delivery.

Defense in depth

35 tenant tables run with FORCE row-level security beneath the application-level checks.

06 Technical decisions

  1. 01Modular monolith over microservices
  2. 02Pure shared policy engine instead of an external authorization service
  3. 03Opaque sessions instead of JWTs
  4. 04Deterministic meeting briefs; no LLM

07 Verified test results

Core (Vitest)
91
API integration
54
Web
16
Playwright
9

Latest GitHub Actions run on main: passing. Counts as reported in the repository's README and testing docs.

Explore the results interactively in the Lab

08 Screenshots

Company Map showing people and teams as a graph
Company Map in People mode
Manager Team Pulse workspace
Manager workspace: Team Pulse
Dependency graph in impact mode
Dependency impact analysis
Roadmap timeline view
Roadmap timeline

Screenshots captured from the running application and published in the repository's docs.

09 Links

No public deployment. Runs locally with Docker Compose. Not publicly deployed.