Track Code Changes and Resume AI Sessions
A skill that tracks code changes as structured JSON records with a state machine, enabling seamless AI session handoff.
Why it matters
Ensure seamless AI coding session continuity by meticulously tracking all code changes. Resume complex tasks with full context, even after session expirations or interruptions.
Outcomes
What it gets done
Initialize and manage structured change records.
Provide session handoff with progress summaries and next steps.
Generate accessible HTML dashboards for change visibility.
Retroactively create change logs from Git history.
Install
Add it to your toolbox
Run in your project directory:
curl -fsSL https://spark.entire.vc/get/ag-technical-change-tracker | bash Overview
Technical Change Tracker
A skill that tracks code changes as structured JSON records through a defined state machine, storing progress/context for seamless AI session handoff, with an accessible HTML dashboard. Use it when a bot session needs to resume mid-task with full context, or when onboarding a project's undocumented change history retroactively from git.
What it does
This skill tracks every code change as structured JSON records with accessible HTML output, ensuring AI coding sessions can resume seamlessly when a previous session expires or is abandoned mid-task. Each tracked change ("TC") moves through a defined state machine: planned to in_progress to implemented to tested to deployed, with a blocked branch off in_progress. It's operated via commands (/tc init, /tc create, /tc update, /tc status, /tc resume, /tc close, /tc export, /tc dashboard, /tc retro). For session handoff, each TC stores a progress summary, next steps, blockers, key context, and the files currently in progress, so the next bot session can pick up exactly where the last one left off rather than re-deriving context from scratch. Bookkeeping runs non-blockingly via background subagents, never interrupting the actual coding work. Structured JSON records maintain an append-only revision history, test cases are stored with log-snippet evidence, and the dashboard output is WCAG AA+ accessible HTML (dark theme, rem-based fonts) built with a CSS-only status-filterable dashboard, using only the Python standard library with zero external dependencies. /tc retro supports retroactively bulk-creating change records from existing git history, useful for onboarding a project with undocumented change history.
When to use - and when NOT to
Use it when structured change tracking is needed across AI coding sessions, when a bot session expires mid-task and the next session needs full context to resume, or when onboarding a project that has undocumented change history and needs retroactive tracking built from git history.
Inputs and outputs
Inputs: code changes being made during a session, plus /tc commands to create, update, or query tracked change records.
Outputs: structured JSON change records with revision history and test evidence, an accessible HTML dashboard with status filters, and session-handoff context (progress, next steps, blockers) for the next AI session to resume from.
/tc create
/tc status
/tc resume
/tc retro
Who it's for
Developers using AI coding agents across multiple sessions who need structured, resumable change tracking so context and progress survive session expiration or handoff between bot sessions.
FAQ
Common questions
Discussion
Questions & comments · 0
Sign In Sign in to leave a comment.