AICHE +IIterm2 Integration
Voice input for Mac terminal
Speak commit messages and docs into iTerm2.
The short answer: open iTerm2, position your cursor at the command prompt or inside a text editor, press ⌃+⌥+R, speak your commit message, documentation, or command description, and AICHE inserts the text directly into the terminal.
iTerm2 is the terminal for macOS power users who outgrew Terminal.app. Split panes, profiles, triggers, tmux integration, search across scrollback. It is where you git commit, ssh into servers, tail logs, and run builds. Most of the text you type in iTerm2 is commands, but a meaningful portion is prose: commit messages, script headers, README content via heredocs, and comments in config files opened with vim or nano.
That prose is where voice helps. Git commit messages in particular suffer from the terminal context. You just finished an hour of focused coding. You stage your changes, type git commit, and now you need to switch from code-thinking to writing-thinking and explain what you did and why. The result is usually "updated auth module" when it should be three sentences of context. Voice captures the full explanation in 15 seconds because you are still in the mental context of the work.
- Open iTerm2.
- Navigate to your project directory and stage your changes with git add.
- Type
git committo open your configured editor (vim, nano, or whatever $EDITOR is set to). - Press ⌃+⌥+R to start AICHE recording.
- Speak your full commit message: what changed, why it changed, and any notes for reviewers.
- Press ⌃+⌥+R again. AICHE inserts the text at your cursor position in the editor.
- Save and close the editor to complete the commit.
Heads-up: AICHE inserts text wherever the cursor is in iTerm2. If you are at the shell prompt, text goes to the prompt. If you are inside vim, nano, or any TUI editor, text inserts at the cursor position within that editor. Make sure you are in the right context before stopping the recording.
Git Commit Messages
Detailed Commit History
The difference between a useful git history and a useless one is commit message quality. git log --oneline tells you nothing when every message is "fix" or "update." Detailed commit messages make git blame valuable and make debugging regressions possible.
In iTerm2, after staging your changes, type git commit to open your editor. Press ⌃+⌥+R and speak the subject line and body. Say "refactor database connection pooling to use HikariCP instead of c3p0. Reduces connection acquisition time from 50ms to 5ms based on load testing with 100 concurrent requests. Updates max pool size to 20 with 30 second timeout. Adds health check query to catch stale connections. Backward compatible, no changes needed in data access layer."
That message takes 20 seconds to speak. It takes 2 minutes to type. Multiply by 10 commits per day, and you save 15-20 minutes of typing while producing a dramatically more useful git history.
Commit Messages with git commit -m
If you prefer inline commit messages, type git commit -m " and then press ⌃+⌥+R. Speak your message, press the hotkey again, then close the quote and press Enter. This works for shorter messages. For multi-line messages, using git commit with an editor gives you more space.
Shell Script Documentation
Script Headers
Every shell script should have a header comment explaining what it does, when to run it, what arguments it accepts, and what it modifies. Open a script in your editor from iTerm2 (vim, nano, or whatever you prefer), position cursor at the top, and dictate the header.
Say "this script provisions a new staging environment by cloning the production database, applying pending migrations, seeding test data, and starting the application server on the specified port. Accepts two arguments: environment name and port number. Environment name must be unique and is used for the database name and log directory. Port must be available. Requires AWS CLI configured with staging credentials and PostgreSQL client installed. Run from the project root directory."
Inline Script Comments
For complex pipeline commands or conditional blocks in scripts, add comments explaining the logic. Press ⌃+⌥+R while your cursor is on a blank line above the command, speak the explanation, then add the # prefix manually.
Heredoc Content
Writing Documentation from the Terminal
Heredocs let you create multi-line text files directly from the terminal. This is useful for generating README files, config templates, or documentation on remote servers where you do not have a GUI editor.
Type your heredoc opening (cat << 'EOF' > README.md), press Enter, then press ⌃+⌥+R and dictate the content. AICHE inserts the text line by line. When finished, type EOF to close the heredoc. The result is a documentation file created entirely by voice.
SSH Session Documentation
When SSH'd into a remote server troubleshooting an issue, you often discover things worth documenting. Open a file with vim or nano on the remote server, press ⌃+⌥+R, and dictate your findings. AICHE inserts into iTerm2, which sends keystrokes to the remote shell, which inserts into the remote editor. No server-side installation needed.
iTerm2 Split Panes for Voice Workflows
Reference While Dictating
iTerm2's split pane feature (Cmd+D for vertical, Cmd+Shift+D for horizontal) lets you keep reference material visible while dictating in another pane. Open your code diff in one pane (git diff --staged) and your commit editor in the other. Scan the diff while speaking your commit message.
For documentation work, open the source code in one pane and a markdown file in vim in the other pane. Read the code in the left pane, dictate the documentation into the right pane. The split layout keeps context visible without window switching.
Log Monitoring During Dictation
When documenting system behavior or writing incident reports, tail a log file in one pane and dictate observations into a file in the other pane. The live log feeds your spoken commentary with real-time evidence.
iTerm2-Specific Tips
- Profile-based workflows. iTerm2 profiles can pre-configure split panes, working directories, and startup commands. Create a "documentation" profile that opens your project directory with a code pane and a documentation pane side by side.
- Search scrollback. After dictating, use iTerm2's search (Cmd+F) to find text in scrollback. This helps verify that commit messages were captured correctly before the editor closes.
- Triggers for feedback. iTerm2 triggers can highlight or alert on specific text patterns. Set a trigger to highlight "TODO" or "FIXME" in your terminal output so dictated comments with those markers stand out.
- tmux integration. If you use tmux through iTerm2's native integration, AICHE works the same way. Text inserts into whatever tmux pane has focus. Voice in tmux sessions over SSH works without any additional configuration.
The pro-tip: Use iTerm2's command history (Cmd+Shift+H) after dictating to review recent text insertions. If a commit message needs editing, you can retrieve it and modify it before the commit goes through.
Result: Git commit messages that explain the "why" behind changes. Shell scripts with headers and inline comments that prevent "what does this do" questions. Documentation files created from the terminal in minutes. The prose tasks in terminal work become as fast as the command tasks.
Do this now: Open iTerm2, navigate to a project with staged changes, type git commit, press ⌃+⌥+R, and dictate a commit message that explains what you changed and why it matters.
Works With
AICHE with GitHub Copilot
GitHub Copilot with voice. Dictate code requests and prompts naturally. Get suggestions without typing.
AICHE with GitHub
Dictate GitHub PR descriptions, issue reports, code review comments, and Discussion answers at speaking speed.
AICHE with JetBrains IDEs
Dictate documentation and comments in JetBrains IDEs. Write code explanations at speaking speed.
AICHE with Linear
Linear issues with voice. Dictate bug reports, features, and updates without typing. Document issues naturally.
AICHE with v0.dev by Vercel
v0.dev UI with voice. Dictate component descriptions and UI requirements naturally. Build interfaces.
AICHE with VS Code
Use AICHE in VS Code for documentation, comments, and README files. Dictate explanations, not code syntax.