Agent Detection & Monitoring

thane automatically detects AI coding agents running in your terminals — Claude Code, Codex, Gemini CLI, Cursor, Aider, GitHub Copilot, Goose, OpenCode, Cline, Amp, Augment, OpenHands, Plandex, Qwen Code, Devin, Tabnine, Cody, and Continue. See active, stalled, or idle status per workspace, with prompt capture and port detection for a complete picture of what your agents are doing.

18 AI Agents Detected

thane recognizes Claude Code, OpenAI Codex, Google Gemini CLI, Cursor, Aider, GitHub Copilot, Goose, OpenCode, Cline, Sourcegraph Amp, Augment Code, OpenHands, Plandex, Qwen Code, Devin, Tabnine, Cody, and Continue. Detection works automatically via process tree inspection — no configuration needed.

Active / Stalled / Idle Status

Each workspace shows its agent status in the sidebar and status bar. Active agents that haven't produced output in 60 seconds are flagged as stalled, helping you spot hung processes.

Prompt Capture

thane captures Claude Code invocations from terminal I/O, extracting the prompt text and distinguishing between interactive and headless (--print) modes.

Port Detection with Sidebar Badges

When your dev server starts listening on a port, thane detects it and shows a clickable badge in the sidebar. Click to open in the embedded browser, Shift+click to open in your default browser.

How it works

thane uses process tree inspection to detect running AI agents. On Linux, it traverses /proc to find known agent binaries spawned from terminal shells. On macOS, it uses libproc for the same purpose. The detection runs periodically and updates the sidebar and status bar in real time.

When Claude Code is detected, thane goes further: it scans terminal I/O for invocation patterns to capture the prompt being executed. This lets you see what Claude Code is working on at a glance, without switching to its terminal.

Port detection monitors listening TCP ports per workspace. On Linux, /proc/net/tcp is parsed; on macOS, lsof provides the same data. Detected ports appear as badges in the sidebar workspace entry, creating a seamless workflow from running a dev server to previewing it in the embedded browser.

Use cases

Multi-agent monitoring

Run Claude Code across multiple workspaces and see at a glance which ones are active, which are idle, and which might be stuck — all from the sidebar.

Dev server workflow

Start your dev server in a terminal, see the port badge appear in the sidebar, click to preview in the embedded browser — all without leaving thane.

Stall detection

Catch agents that have stopped producing output. thane flags active agents as stalled after 60 seconds of silence, so you can investigate or restart.

Ready to try thane?

Free for personal use. OS-level sandboxing, split panes, embedded browser, and real-time audit trail — all on your machine.