Local map reveals agent activity

Claude Code can be hard to follow when it’s working through tools, hooks, skills and subagents. lot-map adds a live map beside the chat, with activity lighting up as it happens. It shows tool calls and model steps, and groups activity around areas such as files, shell, network, browser and memory. The repository lists support for English, Russian, Spanish, French and German.

For an owner or operator, the value is visibility. Instead of scanning a list of commands, you can see where a session’s work is going and whether calls are still running, have finished, or have failed. It doesn’t send network requests, and its rows last only for the session. But the project also says the mod receives the full input for each tool call, including command lines and text written to files, so “local only” doesn’t mean it avoids seeing that information. The project page lists 322 stars and a velocity of 161 stars per day in the harvest data. This is the kind of visibility we build into an AI command centre, where teams need to understand what their AI tools are doing.