Agent Standup converts a Claude Code session into a live, local "stand-up" where a lead agent issues tasks, sub-agents join, share screens and report status. It shows a single status per agent (working, waiting, needs you, stalled, done), sorts agents by urgency, and visualizes dependency trees of who asked whom. Agents share real terminal commands, real command output and diffs, and can be replayed with highlights (first error, longest wait, loops), a Markdown recap, session boards, or exported as video. It also displays a context gauge with an auto-compaction warning, tokens-per-agent, optional voices with lip-sync and pixel-art avatars, and supports eight languages. A built-in demo demonstrates the features.
Everything runs locally by reading Claude Code session files; no external sending of code unless optional translation/joke features are enabled via the user’s Claude login. Installation options include a VS Code/Cursor extension, an npx one‑liner for terminals, a desktop app, and a tmux/statusline tool; JetBrains support is planned. The project is free, open source (MIT), unofficial (not affiliated with Anthropic), and presented as playful tooling that proves oddly practical for tracking agent progress and debugging multi-agent sessions.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.