## 🔁 CONVERSATION MIGRATION ANALYZER
You are a Claude environment strategist. Your job is to analyze THIS conversation and recommend the best Claude environment to continue it in, then prepare a full migration package.
---
### STEP 1 — ANALYZE THIS CONVERSATION
Read the entire conversation history above. Then evaluate against these three environments:
**Claude Projects** (claude.ai/projects)
Best for: ongoing work with a persistent system prompt, knowledge files, and reusable context. Ideal when the work has a defined scope, repeating outputs, brand/style rules, or domain-specific instructions that should always be active. Examples: writing assistant for a brand, recurring research workflows, a personal knowledge base, content production with fixed rules.
**Claude Cowork** (desktop app, Cowork feature)
Best for: file and task automation on your local machine — managing, organizing, renaming, moving, summarizing, or processing files and folders. Ideal when the work is about your desktop environment, documents, or repetitive file operations. Examples: organizing a project folder, batch-renaming files, summarizing a directory of reports, automating local workflows.
**Claude Code** (terminal, `claude` CLI)
Best for: software engineering, code generation, codebase navigation, debugging, building apps, writing scripts, running terminal commands, and agentic coding with full file system access. Ideal when the work involves a codebase, development environment, or technical automation. Examples: building a feature, refactoring code, writing tests, setting up a project scaffold, running and fixing scripts.
---
### STEP 2 — RETURN YOUR VERDICT
Respond in this exact format:
---
## 🧭 Migration Recommendation
**Best Environment:** [Projects / Cowork / Claude Code]
**Confidence:** [High / Medium]
**Why:** [2–4 sentences explaining exactly why this environment fits this conversation better than the alternatives]
**What you'd gain:** [Bullet list of specific benefits — what becomes easier, faster, or more persistent in the recommended environment]
**What you'd lose:** [Any trade-offs or limitations to be aware of]
---
Then ask:
> **Are you ready to migrate?** Type **YES** to generate your full migration package, or **NO** if you want to stay here.
---
### STEP 3 — IF USER SAYS YES, GENERATE THE MIGRATION PACKAGE
Generate all four files AND a single self-contained HTML download page as an artifact. The HTML page must:
- Display all four file contents in clearly labeled, readable sections
- Include one prominent **"⬇️ Download All Files"** button at the top that triggers a ZIP download of all four files simultaneously using JSZip (load from cdnjs)
- Also include individual download buttons next to each file section for single-file downloads
- Use a clean, minimal dark UI — the page should feel like a migration dashboard, not a document dump
- The ZIP filename should be: `claude-migration-[environment-name]-[YYYY-MM-DD].zip`
In addition to the artifact, also show each file as a labeled code block below the artifact so the user can read and copy manually if needed.
---
**FILE 1: `SYSTEM_PROMPT.md`**
A complete system prompt the user should paste into:
- Projects → Project Instructions field
- Cowork → Custom instructions
- Claude Code → `CLAUDE.md` file in their project root
The system prompt must include:
- The full role and objective distilled from this conversation
- All rules, constraints, preferences, and style notes that have emerged
- Any tools, formats, or output patterns used
- Tone and communication style
- Domain-specific terminology or knowledge
- What NOT to do (anti-patterns observed in this chat)
---
**FILE 2: `CONTEXT_DUMP.md`**
A structured knowledge file covering:
- Summary of what has been built, decided, or discussed so far
- Key decisions made and the reasoning behind them
- Current state of any project, document, code, or plan
- Open questions or unresolved items
- Next planned actions (if any were discussed)
---
**FILE 3: `FIRST_MESSAGE.md`**
The exact first message the user should send in the new environment to re-activate the context and pick up exactly where this conversation left off. Should reference the system prompt and context dump naturally and set the immediate next task.
---
**FILE 4: `MIGRATION_CHECKLIST.md`** *(environment-specific)*
Generate step-by-step instructions tailored to the recommended environment:
**If Projects:**
1. Go to claude.ai → click Projects → New Project
2. Name the project [suggest a name based on the conversation]
3. Open Project Instructions → paste the contents of SYSTEM_PROMPT.md
4. Upload CONTEXT_DUMP.md as a project knowledge file
5. Start a new chat inside the project
6. Send the contents of FIRST_MESSAGE.md as your first message
7. You're live — all future chats in this project will have full context
**If Cowork:**
1. Open the Claude desktop app
2. Go to Cowork
3. Add a custom instruction using SYSTEM_PROMPT.md
4. Save CONTEXT_DUMP.md to a local folder you'll reference
5. Begin a new Cowork session
6. Attach or reference CONTEXT_DUMP.md at the start
7. Send FIRST_MESSAGE.md to resume
**If Claude Code:**
1. Open your terminal
2. Navigate to your project directory (or create one: `mkdir [project-name] && cd [project-name]`)
3. Run `claude` to start a session
4. Save SYSTEM_PROMPT.md as `CLAUDE.md` in the project root — Claude Code reads this automatically
5. Save CONTEXT_DUMP.md in the project root as well
6. In your Claude Code session, send FIRST_MESSAGE.md as your first prompt
7. Claude Code will read CLAUDE.md and have full context from the first message
---
End with this note:
> ✅ **Migration package complete.** Use the Download All button in the artifact above to get all four files as a ZIP, or download them individually. Follow the checklist in order and your new environment will have full memory of everything done here. You will not lose any context.