🧠 AI Prompt Versioning & Continuation System

A structured protocol for maintaining continuity, context, and version control across AI chat sessions.

This system solves one of the most frustrating problems with AI chat tools: context death. Every new chat starts cold. Work, decisions, code patterns, and agreements made in previous sessions vanish. This prompt protocol is designed to eliminate that problem by establishing a repeatable, structured handoff system between sessions.

The core idea is simple: at any point in a working session, you issue a single command and the AI generates everything needed to reconstruct full context in a new chat — no re-explaining, no re-litigating decisions, no starting over.

• AI chats have no memory across sessions
• Complex, multi-session projects lose critical context
• Re-explaining background wastes time and degrades output quality
• No standard format for "picking up where we left off"
• No audit trail of decisions made, patterns established, or next steps agreed upon

Issue any of the following phrases to activate the versioning system:

When a version update is triggered, the AI produces four artifacts:

1. Summary — ~450 words covering what was accomplished, with delta notes from the previous version
2. Change Log — Downloadable in both .txt and .html format, with timestamps and version numbers
3. Continuation Prompt — A .txt file containing the full prompt needed to re-open a new chat with complete context
4. Helper Files — Supporting context in JSON, HTML, or TXT format (whichever is most efficient) for the AI to parse rapidly on re-entry

Note: The prompt regeneration is never included in a normal response. It is only produced on an explicit Update Version request.

/session-exports/
├── v1.0_summary.txt          # Human-readable session summary
├── v1.0_changelog.html       # Formatted changelog with versioning
├── v1.0_changelog.txt        # Plain text changelog
├── v1.0_continuation.txt     # The re-entry prompt for next session
└── v1.0_context.json         # Machine-readable helper context

Each version entry in the changelog includes:

• Version number (e.g., v1.0, v1.1, v2.0)
• Date/time of the update (when available)
• Summary of the session at the top
• Key decisions made
• Code patterns established
• Next steps identified
• Suggestions agreed upon
• Any additions or changes from the prior version

To resume a session in a new chat:

1. Open the saved v[X.X]_continuation.txt file
2. Paste its full contents as your first message in the new chat
3. Attach any relevant v[X.X]_context.json helper files if referenced
4. The AI will confirm it has absorbed the context and is ready to continue

• Speed — JSON helper files allow faster AI parsing vs. long prose prompts
• Fidelity — Detailed changelogs prevent decision drift across sessions
• Portability — Plain text formats work across any AI tool or interface
• Auditability — Every version is a recoverable snapshot of the project state

v1.0 — Initial system definition. Project scope, structure, and trigger commands established. Underlying project (referenced in the prompt body) is pending scoping questions before first working session begins.

• Ask clarifying questions before completing a version update if context is ambiguous
• Helper files can be in any format — use whatever is most machine-readable
• The system is tool-agnostic: works with Claude, GPT, Gemini, or any chat-based AI
• This README.md itself should be versioned alongside the prompt files

Built for multi-session AI workflows where context continuity is non-negotiable.