thinkt#
Tools for AI assistant session exploration and extraction.
thinkt provides tools for exploring and extracting data from AI coding assistant sessions.
Supported Sources#
We created a common thinkt interface to enable uniform access to various Sources. We maintain a library of implementations and currently support:
claudeClaude Code from AnthropickimiKimi Code from MoonshotgeminiGemini CLI from GooglecopilotCopilot CLI from GitHubcodexCodex CLI from OpenAIqwenQwen Code from Alibaba
Quick Start#
# Launch the interactive TUI, depending on context
# If in a project folder, it will open a session view
thinkt
# List all projects
thinkt projects
# View sessions for a project
thinkt sessions
# Open the Web interface
thinkt web
# Search across sessions
thinkt search "authentication"Features#
We have several ways to explore your conversations:#
- Interactive TUI - Navigate projects, sessions, and conversations with a keyboard-driven terminal UI
- Web Interface - Full webapp for visual trace exploration via
thinkt web - MCP Server - Expose session data via Model Context Protocol for AI assistants
- REST API - Programmatic access with OpenAPI/Swagger documentation
All with the same shared features:#
- Multi-Source Support - Auto-discovers Claude Code, Kimi Code, Gemini CLI, Copilot CLI, Codex CLI, and Qwen Code sessions
- Full-Text Search - Search across all indexed sessions via
thinkt search - Semantic Search - Find sessions by meaning with on-device embeddings via
thinkt semantic search - Prompt Extraction - Extract user prompts in markdown, JSON, or plain text
The thinkt platform is meant to make your work easier:#
- Themes - 14 built-in color schemes (Dracula, Nord, Catppuccin, etc.) with interactive browser and iTerm2 import
- Languages - Support for English, Chinese, and Spanish with easy switching
- App Management - Configure open-in apps and default terminal via
thinkt apps - Embedding Management - Configure and manage embedding models via
thinkt embeddings
With operational tools:#
- Docker - Container images for sandboxed deployment
- Agent Teams - Inspect multi-agent teams (Claude Code), including members, tasks, and messages
- Prometheus Metrics - For server, indexer, and collect monitoring
- Lite Server - Lightweight debug interface for quick inspection
- Agent Teams - Inspect multi-agent teams (Claude Code), including members, tasks, and messages
Installation#
# Using Go
go install github.com/wethinkt/go-thinkt/cmd/thinkt@latest
# Using Docker (sandboxed, read-only access)
docker pull ghcr.io/wethinkt/thinkt:latestSee the Docker Guide for sandboxed usage with volume mounts.
Next Steps#
- Read the CLI Guide for common workflows and examples
- Configure the MCP Server for AI assistant integration
- Browse the Command Reference for detailed usage of all commands
- Run with Docker for sandboxed, read-only access
- Set up Trace Collection for multi-machine trace aggregation
- Explore the REST API with interactive Swagger documentation
- See the LLM Guide for AI assistant integration tips
- View the project on GitHub