old-claude-mcp

# old-claude-mcp [](LICENSE) [](https://github.com/sneg55/old-claude-mcp/stargazers)    Anthropic lets you export your Claude conversation history but provides no way to import it into a new account. This is a local MCP server that makes the export actually useful — index your old conversations into SQLite and Claude can search them automatically in new sessions. **Background:** [Claude lets you export your data but won't let you use it](https://sawinyh.com/blog/claude-export-no-import) ## Why If you switch Claude accounts, lose access to an email address, or just want your old context available going forward, you're stuck with a JSON file you can't do anything with. This tool indexes that file and exposes it as three MCP tools Claude calls automatically during conversations. ## Setup **1. Export your Claude history** Go to claude.ai → Settings → Export Data. You'll get a ZIP containing `conversations.json`. **2. Install dependencies** ```bash git clone https://github.com/sneg55/old-claude-mcp cd old-claude-mcp pip install -r requirements.txt ``` **3. Index your export** ```bash python indexer.py ~/Downloads/conversations.json --db history.db ``` **4. Add to Claude Desktop config** Edit `~/Library/Application Support/Claude/claude_desktop_config.json`: ```json { "mcpServers": { "chat-history": { "command": "/usr/local/bin/python3", "args": [ "/path/to/old-claude-mcp/server.py", "--db", "/path/to/old-claude-mcp/history.db" ] } } } ``` Use an absolute path to your Python interpreter. Find it with `which python3`. **5. Restart Claude Desktop** ## Usage Just talk normally. Claude calls the tools automatically when relevant: - "What did we decide about X?" → `search_history("X")` → `get_conversation(uuid)` - "Load context for project Y" → `search_history("Y")` - "Find all conversations about company Z" → `search_history("Z")` ## Tools | Tool | Description | |------|-------------| | `search_history(query, limit?)` | Full-text search across all conversations, returns ranked results with snippets | | `get_conversation(uuid)` | Full message thread for a conversation (truncated at 50k chars) | | `list_conversations(limit?, after?)` | Browse conversations by date descending | ## Adding New Providers Implement `ConversationParser` from `parsers/base.py` and pass it to `insert_conversations()` in `indexer.py`. ChatGPT export support (`ChatGPTParser`) is a natural next step — PRs welcome. ## Future - Semantic/vector search (ChromaDB + sentence-transformers) - ChatGPT export support - Incremental indexing for updated exports