emacs-claude-code: Intelligent Claude Integration for Emacs
Yusuke Watanabe (he/him) - Pronunciation: you-SKAY wah-tah-NAH-bay, GitHub: https://github.com/ywatanabe1989/emacs-claude-code, ywatanabe@alumni.u-tokyo.ac.jp
The following image shows where the talk is in the schedule for Sat 2025-12-06. Solid lines show talks with Q&A via BigBlueButton. Dashed lines show talks with Q&A via IRC or Etherpad.
Format: 20-min talk ; Q&A: ask questions via Etherpad/IRC; we'll e-mail the speaker and post answers on this wiki page after the conference
Discuss on IRC: #emacsconf
Status: Waiting for video from speaker
Saturday, Dec 6 2025, ~11:40 AM - 12:00 PM MST (US/Mountain)
Saturday, Dec 6 2025, ~10:40 AM - 11:00 AM PST (US/Pacific)
Saturday, Dec 6 2025, ~6:40 PM - 7:00 PM UTC
Saturday, Dec 6 2025, ~7:40 PM - 8:00 PM CET (Europe/Paris)
Saturday, Dec 6 2025, ~8:40 PM - 9:00 PM EET (Europe/Athens)
Sunday, Dec 7 2025, ~12:10 AM - 12:30 AM IST (Asia/Kolkata)
Sunday, Dec 7 2025, ~2:40 AM - 3:00 AM +08 (Asia/Singapore)
Sunday, Dec 7 2025, ~3:40 AM - 4:00 AM JST (Asia/Tokyo)
Description
As AI coding assistants become essential development tools, integrating them seamlessly into our Emacs workflows becomes crucial. This talk introduces emacs-claude-code, a package that transforms how Emacs users interact with Claude Code through intelligent automation and enhanced vterm integration.
The main challenge with terminal-based AI assistants is the repetitive manual interaction required. Claude Code presents various prompts (Y/N choices, waiting states, continuation prompts) that interrupt the development flow. emacs-claude-code solves this by introducing smart auto-response patterns that recognize Claude's state and respond appropriately, allowing developers to maintain focus on their code rather than managing the AI interface.
Key features I'll demonstrate include:
Intelligent Auto-Response System: Automatically handles Claude's various prompt states (INITIAL WAITING, Y/N, Y/Y/N, WAITING) with customizable responses. This allows uninterrupted AI-assisted coding sessions.
Centralized Buffer Management: A dashboard view of all Claude sessions showing their current state, auto-response status, and last interaction time. Users can quickly navigate between multiple AI conversations and toggle automation settings.
Yank-as-File Functionality: Elegantly handles large code snippets by saving them as files instead of cluttering the terminal, with full TRAMP support for remote development.
Periodic Command Automation: Executes custom commands at specified interaction intervals, perfect for maintaining context or triggering regular actions like compacting conversation history.
The package leverages Emacs' powerful vterm-mode to create a robust integration layer. I'll share implementation details about state detection using regular expressions, buffer management techniques, and the advice system used for seamless clipboard integration.
Beyond the technical implementation, I'll discuss practical workflows that emerge from this integration, including custom Claude commands (stored as Markdown files) that can be triggered automatically based on context. The talk will include a live demonstration showing how these features work together to create a fluid AI-assisted development experience.
This project embodies the Emacs philosophy of extensibility and automation, making cutting-edge AI tools work the way Emacs users expect - efficiently, automatically, and under their complete control.
About the speaker:
I'm Yusuke Watanabe, a researcher who uses Emacs for both academic work and software development. After experiencing friction with manual Claude Code interactions, I developed emacs-claude-code to automate repetitive tasks and integrate AI assistance smoothly into Emacs workflows. This project represents my approach to making modern AI tools work within the Emacs ecosystem while maintaining the editor's core principles of efficiency and user control.
Questions or comments? Please e-mail ywata1989@gmail.com