Overview
MCP Toolz is an MCP server for Claude Code that enables persistent context and todo management across sessions with multi-model AI feedback. It exposes MCP tools such as context_save, context_search, context_get, context_list, and context_delete to save contextual artifacts like conversations, code snippets, architectural ideas, or error traces. It also provides todo tools (todo_save, todo_list, todo_restore, etc.) to snapshot and restore todos. Contexts saved during a Claude Code session are automatically tagged with a unique session_id, a session_timestamp, and the current project_path for easy restoration and cross-session tracking. Data is stored in a single SQLite database (~/.mcp-toolz/contexts.db) and can be configured to a synced location via MCP_TOOLZ_DB_PATH. It supports discovering MCP resources passively (contexts and todos via mcp-toolz:// URIs). The server works with Claude Code and can be used via CLI or Claude Code MCP settings. Requires API keys for AI backends (OpenAI, Anthropic, Google, DeepSeek) via environment variables or .env. Provides a convenient way to save contexts across projects and receive AI feedback from multiple models (ChatGPT, Claude, Gemini, DeepSeek).
Features
MCP Server integration with Claude Code
Provides full MCP tool integration ready to use inside Claude Code, enabling direct access to context and todo commands.
Session Continuity
Restores last session contexts when Claude Code restarts, preserving work state.
Project Organization
Organizes contexts and todos by project directory for easy navigation and management.
Session Tracking
Each Claude Code session gets a unique ID; contexts are tagged with session_id and session_timestamp.
AI Feedback across multiple models
Get insights from ChatGPT, Claude, Gemini, and DeepSeek on your contexts and decisions.
Context Types
Supports saving conversations, code snippets, architectural ideas, or error traces.
Persistent Todos
Save and restore your todo lists across sessions with per-project organization.
Full-Text Search
Find contexts by content, tags, project, or session across the stored database.
