botctl

botctl is a process manager for autonomous AI agent bots, letting you run persistent agents from a single CLI with a terminal dashboard (TUI), web UI, and declarative configuration. Each bot is defined by a BOT.md file with YAML frontmatter and a markdown system prompt, making configuration version-controllable and human-readable. Bots cycle through running, sleeping, paused, and stopped states, with support for message queues, session preservation, and skill discovery. It is written in Go and ships pre-built binaries for macOS, Linux, and Windows.

• TUI Dashboard — launch botctl with no arguments to open an interactive terminal dashboard for managing all bots; navigate with vim-style keys.
• Web UI — run botctl --web-ui to open a browser-based dashboard on port 4444 (or a custom port).
• Declarative BOT.md Config — define each bot's name, interval, max turns, workspace, skills directory, and system prompt in a single markdown file with YAML frontmatter.
• Pause / Play / Stop — pause a running bot to preserve its Claude session context, then resume with botctl play; stop kills the process entirely.
• Skills System — install community or custom skills from GitHub repos via botctl skills add; skills are discovered from global, botctl-wide, and per-bot directories.
• Message Queue — send a message to a running or sleeping bot with botctl start my-bot --message "..." to inject tasks mid-cycle.
• Log Management — view and follow per-run log files with botctl logs my-bot -f; configure retention count per bot.
• Self-Update — run botctl update to fetch and replace the binary in-place with the latest release.
• SQLite State Store — all bot state, stats (cost, turns, session), and metadata are persisted in a local SQLite database at ~/.botctl/botctl.db.
• Easy Install — install via a one-liner curl script or go install github.com/montanaflynn/botctl@latest.