Observability UI AI SDLC Harness

AI-driven software development lifecycle harness for the Observability UI team. This repo provides structured context, task tracking, and automation for using AI coding agents across the team's project portfolio.

How it works

Each task follows a three-document workflow:

1. spec.md - Problem statement, related projects/branches, and acceptance criteria.
2. plan.md - Step-by-step breakdown an AI agent can execute against.
3. execution.md - Progress tracking with checkboxes and notes captured during execution.

Tasks live in tasks/. The projects/ directory contains git submodules for every repo in scope, giving agents direct access to source code.

Projects

See ARCHITECTURE.md for the full project catalog and system architecture.

Repository layout

tasks/                  # Active tasks (description + work-plan + execution)
completed/              # Archived completed tasks
projects/               # Git submodules for all in-scope repos
bin/                    # Local tooling (dprint)
claude/plugins/obsui/   # Claude Code plugin for assisted development and code reviews

Setup

git clone --recurse-submodules https://github.com/observability-ui/harness/
make setup    # install tools and reset submodules to their configured branches

Resetting projects

After working on tasks, submodules may have checked-out branches or uncommitted changes. Run make reset-projects to reset all submodules back to the branches defined in .gitmodules at the latest remote HEAD. This prevents intermediate states from being committed to this meta-repo.

Markdown formatting

All markdown is formatted with dprint (150 char line width). Run make lint to format and make check to validate.