What is whatsmyagentdoing?
whatsmyagentdoing is a local dashboard for keeping track of coding-agent sessions running on your machine. It brings activity from supported command-line agents into a small, animated office that opens in a browser, with a panel for session details and recent conversation. It is aimed at developers who run multiple agents or sessions and want to see which are working, waiting for input, or finished without checking each one separately.
whatsmyagentdoing Features:
Agent Support: It reads sessions from Claude Code, Codex, Gemini CLI, Copilot CLI, OpenCode, and other supported coding agents.
Live Status: The dashboard shows whether a session is thinking, running a tool, writing a reply, waiting for you, done, interrupted, or in error.
Office View: Agents appear at desks in a 2D or 3D office, with activity cues such as a thought bubble or code scrolling on a monitor.
Session Details: A side panel groups sessions by whether they need you, are working, or are finished, and shows model, folder, context usage, and recent conversation.
Alerts and Stalls: Browser notifications and a chime can flag agents that need attention, while extended tool runs or silent thinking are marked as no progress.
Local Usage: A daily view summarizes turns and tokens by agent and project, and a private mode hides prompts, replies, and session titles.
whatsmyagentdoing Benefits:
Fewer Check-Ins: A shared overview makes it easier to monitor several sessions without repeatedly opening each agent's interface.
Faster Follow-Up: Seeing which sessions need input helps you prioritize questions and approvals before other work stalls.
Earlier Intervention: No-progress flags can bring a potentially stalled command or session to your attention before it goes unnoticed.
Clearer Usage: Daily totals by agent and project help you understand where your local agent activity is concentrated.
More Privacy: Read-only local access and private mode let you monitor activity without sending session content away from your machine.
Use Cases:
Multi-Agent Monitoring: Keep an eye on several coding agents running at once and check their status from one browser tab.
Approval Follow-Up: Use the needs-you grouping and alerts to spot sessions that are waiting for a question or approval.
Long Task Checks: Watch for a tool call that has run for an extended period or a thinking session that has gone quiet.
Session Handoffs: Review a session's recent conversation and copy its resume command when returning to work on that agent.
Daily Usage Review: Check turns and tokens by project and agent when reviewing a day's coding-agent activity.

