agent-deck and lazyagent
These are competitors—both provide unified terminal interfaces for monitoring and managing multiple AI coding agents, with overlapping support for Claude and other LLMs, so users would typically choose one TUI solution rather than run both simultaneously.
About agent-deck
asheshgoplani/agent-deck
Terminal session manager for AI coding agents. One TUI for Claude, Gemini, OpenCode, Codex, and more.
Built in Go with a stdio-based agent transport protocol, it provides real-time status detection (running/waiting/idle/error) and supports MCP socket pooling to reduce memory overhead by 85-90% across multiple concurrent sessions. Integrates deeply with Claude Code, OpenCode, and standard LLM APIs while offering advanced features like git worktree isolation, Docker sandboxing per session, and a Conductor system for cross-session orchestration.
About lazyagent
illegalstudio/lazyagent
Monitor all your coding agents from one terminal - Claude Code, Cursor, OpenCode, pi and more
Provides three interfaces (terminal UI, macOS menu bar app, HTTP API) for passively monitoring AI coding agent sessions by parsing local session files—JSONL from Claude Code/Codex/pi and SQLite from Cursor/OpenCode—without requiring agent modifications or external servers. Detects real-time activity states (thinking, waiting, reading, writing, etc.) and surfaces detailed session metadata including git context, token usage, cost estimation, and conversation history through a shared state machine architecture.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work