Hi everyone, I'm Shahbaz. This is my first post here, so I'll say who I am
and what I'm working on.
What I build
I'm the developer behind Nusk, a project management
tool I've been building for a while. The idea is simple: most teams run
one project across four or five tools. Tasks in Jira, the schedule in MS
Project or Smartsheet, documentation in Confluence, risks in a
spreadsheet. Each tool is fine alone. Together they drift apart, and
someone ends up copying updates between them.
Nusk puts these in one workspace:
- Work items, boards and sprints, with child items and custom fields
- Gantt scheduling with dependencies and critical path
- A spreadsheet-style grid with formulas
- A wiki, plus RAID logs (risks, assumptions, issues, dependencies)
- Meeting minutes, timesheets, and portfolio-level reports
- Importers for Jira, Smartsheet and MS Project files
- Support for SAFe planning
It runs as a hosted service, and there is a self-hosted option for teams
that need it on their own servers.
The part that surprised me
Nusk has a built-in MCP server. You can point Claude Code, Cursor or
Codex at it, and the agent can read and update your real project data:
list your open work, create a task, check a sprint, read a risk log.
Changes made this way are labeled in the revision history and can be
reverted. I'll write a full walkthrough of the setup in a later post.
What I'll write about here
I'll mostly cover things I run into while building it:
- Designing permissions that hold up on every endpoint
- Importing messy data from other tools
- Building a scheduling engine
- Giving AI agents access to a real system safely
I'll include what went wrong, since that's usually more useful than what
went right.
Try it
There's a demo that needs no signup: nusk.dev/demo
creates a sample workspace that lasts an hour. If you try it, I'd like to
hear what's confusing or missing. Comments here work, and so do direct
messages.
Looking forward to reading what you all are building.
