AI-SDLC

2025shipped

Markdown CLI for 8-step AI development workflows. Everything is just a prompt.

View project

// GitHub

View Repository

51 commits

Last commit 1 months ago

Python

// Problem

Developers skip planning with AI and jump straight to code, then wonder why results are chaotic. Existing tools lock you into providers or hide the workflow.

// Solution

An 8-step CLI: idea → PRD → architecture → patterns → tasks → tests. Each step outputs markdown that feeds the next. Generates prompts for any AI—no API calls, no vendor lock-in.

// What I Built

Python CLI on PyPI. 'aisdlc init' creates templates. 'aisdlc next' merges output with the next template into a ready-to-paste prompt. Chat mode (steps 1-5) for refinement, Agent mode (steps 7-8) for autonomous execution. All artifacts are diffable markdown in git.

// Technologies

Python + uv

Python 3.11+ with zero dependencies beyond stdlib.

TOML + JSON State Management

Config in TOML, state in JSON. Human-readable and git-friendly.

Markdown-Driven Prompts

8 templates with placeholder injection for context chaining.

PyPI Distribution

Install via 'pip install ai-sdlc'. Versioned releases.

CLI Architecture

Subcommands: init, new, next, status, done. Validated state transitions.

// Lessons Learned

01Everything is just a prompt. Requirements, architecture, tasks—all prompts with different contexts.
02AI-SDLC is a linear harness: output from step N becomes input to step N+1.
03Forced structure prevents vibe coding. The 8-step gate makes you plan before you code.
04Tool-agnostic beats tool-specific. Generating prompts means instant model switching.

← Back to all projects