Traceable Searchable ADR Memory Index

What it is

A CLI framework installed into existing projects to provide structured session tracking, Architecture Decision Records (ADRs), and parallel AI agent sprints using git worktrees. It separates framework logic from project data, ensuring all sprint briefs, sessions, and decisions are committed to the project repo for full git history traceability.

Features

• Automates parallel AI agent sprints using isolated git worktrees
• Enforces structured Architecture Decision Records (ADRs) and session logs
• Provides a web dashboard for tracking sprint lifecycle and pipeline status
• Ensures full traceability by committing all memory artifacts to project git history
• Includes validation and auto-fix scripts for sprint brief formatting
• Supports multi-agent coordination via tmux sessions and shared context

Quickstart

./toolkit/setup.sh /path/to/your/project
vim .sprint/config.sh
vim SPRINT_BRIEF.md
./.sprint/scripts/sprint-run.sh
python3 dashboard/server.py

Architecture

flowchart TD
    User[User] -->|Runs| Setup[toolkit/setup.sh]
    Setup -->|Scaffolds| Project[Target Project Repo]
    Project -->|Contains| Scripts[.sprint/scripts/]
    Project -->|Contains| Memory[docs/project-memory/]
    Project -->|Contains| Brief[SPRINT_BRIEF.md]
    Scripts -->|Executes| Agents[Claude Code Agents]
    Agents -->|Work in| Worktrees[Git Worktrees]
    Worktrees -->|Merge Results| MainBranch[Main Branch]
    Dashboard[Dashboard Server] -->|Reads| Memory
    Dashboard -->|Reads| Brief
    Dashboard -->|Visualizes| User

How it's built

The framework consists of Bash scripts for sprint orchestration (init, run, merge, validate) and a Python/HTML dashboard for visualization. It includes an optional TypeScript-based 'Afterburner' agent engine that manages context, skills, and tool execution for autonomous coding agents. Setup is handled via a shell script that scaffolds the target project with necessary directories and configuration files.

How it runs

sequenceDiagram
    participant U as User
    participant S as Sprint Scripts
    participant G as Git Worktrees
    participant A as AI Agents
    participant M as Project Memory
    
    U->>S: Execute sprint-run.sh
    S->>M: Archive current SPRINT_BRIEF.md
    S->>G: Create isolated worktrees
    S->>A: Launch agents in tmux
    A->>G: Read code & context
    A->>A: Execute tasks & generate code
    A->>G: Commit changes in worktree
    S->>G: Validate & Merge worktrees
    S->>M: Save session docs & ADRs
    S->>U: Report completion

How to apply & reuse

Run the setup script against your target project directory to install sprint scripts and templates. Configure test commands and ephemeral files in the generated config. Use the provided scripts to manage sprint lifecycles, or integrate the TypeScript agent engine for automated task execution within isolated git worktrees.

At a glance