THE ORIGIN: DevUssy is the most-starred project in the Ussyverse (15 stars) and the methodology that powers Geoffrussy's planning engine.
The Circular Development Pipeline.
DevUssy is a methodology and toolkit for generating development plans that are reusable, portable, and agent-agnostic. Run an adaptive interview, analyze complexity, generate a plan, hand it off to any AI agent or human developer. The plan is a Markdown artifact — no database, no vendor lock-in, no runtime state.
Guided Q&A captures goals, constraints, target stack, and edge cases. Adapts questions based on your answers.
Python-based scoring rubrics rate difficulty and estimate phase count based on feature density and technical scope.
Frontier models combine interview data with complexity scores. Small projects stay minimal; large ones get detailed architecture.
Rule-based checks plus automated review catch conflicting requirements or hallucinated components before planning begins.
Generates dynamic phases with explicit tests and acceptance criteria in standard Markdown. Streamed live via Server-Sent Events.
Exports state for agent handoffs. The output of the last cycle feeds back as context for the next iteration.
Most development pipelines are linear: Plan > Code > Deploy. DevUssy acknowledges that software is never "done." The Circular Handoff stage feeds the output of the last cycle back into the context for the next iteration. Project state is preserved in files, not in an agent's memory.
DevUssy can be used standalone as a Python CLI (python -m src.cli interactive), or through Geoffrussy where it powers the planning engine. The methodology is the same either way — DevUssy handles the thinking (interview, complexity, design, validation, plan generation) while Geoffrussy adds the doing (file I/O, git, terminal, code execution).
DevUssy is at v0.4.0. The core pipeline works end-to-end. Some reiteration phases in the LLM correction loop are still being refined. It ships with three visual themes (Bliss/Windows XP-style, Terminal/Matrix-style, and Default) and live streaming progress.
git clone https://github.com/mojomast/devussy.git cd devussy && pip install -e . python -m src.cli interactive