Continue V2

Continue V2 — Cross-Tool Workspace Creation

What It Does

Continue V2 is Memora's core differentiating feature. It aggregates conversations, git history, and project files from multiple AI tools and sessions into a new workspace directory — ready to use with any AI coding tool.

This is what no single AI tool can do on its own: cross-tool, cross-session, cross-time context restoration.

How It Works

Step 1: Select Content

Click Continue on any project to open the wizard. Choose what to include:

• Conversations: Select which sessions to include (from any tool)
• Project Files: CLAUDE.md, AGENTS.md, design docs, etc. (grouped by knowledge layer)
• Git History: Recent 20, 50, or full commit history
• Blackboard: AI-generated project analysis
• Target directory: Where to create the workspace

An estimated context file size updates as you select/deselect items.

Step 2: Create Workspace

Memora performs:

1. git clone of your codebase to the target directory
2. Generates MEMORA_CONTEXT.md with all selected content:
   • Project status (Blackboard summary)
   • Git commit history
   • Full conversation transcripts (cleaned, formatted)
   • Project file snapshots
3. Appends a reference to CLAUDE.md / AGENTS.md so AI tools auto-read the context
4. Registers the new directory as a Memora project

Step 3: Use It

Open the workspace with any AI coding tool. The tool reads MEMORA_CONTEXT.md and has your full project context — conversations, decisions, and code history.

MEMORA_CONTEXT.md Structure

# Memora Context: {project name}

> Generated by Memora at {timestamp}
> Source: {N sessions, M commits, K files}

## 1. Project Status
{Blackboard summary}

## 2. Git History (Recent Outcomes)
- `abc1234` fix: resolve auth middleware issue
- `def5678` feat: add user dashboard
...

## 3. Conversations
### Claude Code · Mar 31 15:24
**YOU** (15:24)
Help me refactor the auth module...

**AI** (15:25)
I'll start by reviewing the current implementation...
...

## 4. Project Files Snapshot
### CLAUDE.md
{content}

## 5. Metadata

Design Principles

• Raw conversations, no compression: AI tools have large context windows. Let them read the full conversation rather than a lossy summary.
• User controls scope: You choose what to include. Not everything is dumped.
• Standard Markdown: MEMORA_CONTEXT.md works with any tool. No Memora dependency.
• Cleaned content: Internal tags (<system-reminder>, <local-command-stdout>, etc.) are stripped.