Rogasper
  • Home
  • Projects
  • Blog
  • Course
  • Contact
Hire me
Hire me
All projects
Open source

Onesist

Open-source AI workspace for System Analysts — turn FSD documents into API specs, ERD/DBML, tasks, SIT cases & Gantt timelines. Built-in chat agent, local CLI agents (OpenCode, Claude Code, Codex) and external MCP servers. Desktop (Tauri 2) + Web.

Visit website Source code
React 19
TanStack Start
Tauri 2
Bun
Drizzle ORM
SQLite
Onesist screenshot
  • Overview
  • Features
  • Challenge
  • Outcome
  • Stack
Overview

Why it exists

Problem

System Analysts spend weeks manually converting FSD documents into API contracts, ERD schemas, task breakdowns and traceability matrices — slow, error-prone, and disconnected.

Who it's for

System Analysts, product teams, and technical writers who need to ship handoff-ready specs fast.

Approach

A full-stack desktop & web workspace that delegates FSD → spec conversion to local CLI AI agents, with interactive ERD studio, OpenAPI viewer, phase-grouped tasks, SIT manager, RTM, and Excalidraw canvas — all in one flow.

Features

What it does

Next project
Reurci

AI-powered CV tailoring assistant — paste a job description, get 3 summary variants, paraphrased experiences, ATS score & PDF export. Human-in-the-loop, privacy-first. In development.

Work with me
Need somethinglike this built?
Let's talk
RogasperRogasper

Personal site of Royan Gagas — building notes, courses, and blog sharing what I learn about engineering, SaaS & AI.

Banyumas, Indonesia

contact: [email protected]

Quick Links

  • Blog
  • Course
  • Contact
  • Privacy Policy
  • Terms of Service

Connect

© 2026 Rogasper. All rights reserved.

01
FSD analyzer: rich MD editor, PDF/DOCX → Markdown, one-click AI run
  • 02
    ERD Studio: DBML visual editor with ReactFlow + Dagre, SQL/DBML export
  • 03
    API Specs viewer: OpenAPI parsing, module navigation, payload previews
  • 04
    Tasks & phases: story points, bulk actions, Jira/Monday export, Gantt timeline
  • 05
    Traceability Matrix (BR → FR → DS → TC) with AI gap detection
  • 06
    Native chat agent with BYOK providers, per-thread workspaces and auto-summarised context
  • 07
    External MCP servers (GitHub, Figma, Miro, Monday, custom) with per-server read-only approval gates
  • 08
    Embedded agent terminal (xterm.js + WebSocket) — OpenCode / Claude / Codex / Antigravity
  • 09
    Desktop (Tauri 2, macOS + Windows) + Web — sidecar engine, tray, auto-updater
    • The hard part
    Bridging a real SA pain point: making AI-generated artifacts trustworthy, editable, and traceable — not a black-box demo.
    Outcome

    Where it ended up

    Built from my own SA work at SEVA. MIT-licensed on GitHub with 30+ stars, 5 forks and steady releases (v0.4.2, Oct 2026).

    Stack

    Built with

    React 19
    TanStack Start
    Tauri 2
    Bun
    Drizzle ORM
    SQLite