Disciplined-AI-Software-Development
Disciplined AI Collaboration: a method for building software with AI. One loop at every size, rules held by checks rather than attention, state derived rather t
What is Disciplined-AI-Software-Development?
Disciplined AI Collaboration: a method for building software with AI. One loop at every size, rules held by checks rather than attention, state derived rather than written, evidence in place of claims. Includes the PAG grammar, an architecture and a queryable ontology. Every page is also served as Markdown and JSON at banes-lab.com.
Overview of Disciplined-AI-Software-Development
What it is
Disciplined Methodology is a method for building software with an LLM that writes most of the code. It replaces reminders with automated checks, chat history with files in the project, and confidence with evidence. The method applies one loop of ten steps to every piece of work, from a one-line fix to a plan with several phases. It includes the PAG grammar, an architecture and a queryable ontology. Every page of the site is also served as Markdown and JSON at banes-lab.com.
Key features
- One loop of ten steps applied at every scale, from a single fix to a multi-phase plan - Rules held by automated checks rather than human attention - State derived from files rather than written by hand - Evidence in place of claims for verification - PAG (Pattern Abstract Grammar) for structured instructions to LLMs - Checks written against patterns rather than particular libraries - Every fact declared in one place
Use cases
- Building software with an LLM that writes most of the code - Long projects where context dilution and architecture drift are common - Projects requiring automated enforcement of rules and constraints - Collaboration between developer and model with clear division of work - Scaling a project once its structure is predictable
Getting started
See the repository README for setup instructions.
License and pricing
Licensed under CC-BY-SA-4.0. Open source and free.
Based on the project's public README and repository metadata.