I
Tuned for IDEs
Enhanced dictation knows when the focused app is VS Code, Cursor, Zed, or JetBrains. It applies a code-aware polish profile — more code-fenced spans, fewer overly conversational rewrites.
→ App-specific polish profiles
Code review comments at 11pm. Commit messages on a walk. Voice notes that land directly in the file you’re editing — and a CLI, REST, and MCP server that let your scripts and agents read every transcript Dettivo has ever captured.
§ The current circuit
You start by pasting into ChatGPT for grammar passes. You move to native macOS dictation, which gives up on punctuation. You hand-edit every PR comment. You context-switch out of the editor to type a message, then back in. None of this is captured anywhere your agent can see.
I
Enhanced dictation knows when the focused app is VS Code, Cursor, Zed, or JetBrains. It applies a code-aware polish profile — more code-fenced spans, fewer overly conversational rewrites.
→ App-specific polish profiles
II
Dettivo ships a stdio MCP server. One config line drops it into Claude Code, Cursor, or Codex. Your agent gets tools to search transcripts, fetch the latest, insert one into the editor.
→ MCP server
III
dettivo status, dictation start/stop, transcript latest --format markdown, mcp serve. Pipe transcripts into anything. Script anything. Exit codes you can branch on.
→ CLI tool
IV
One toggle in Settings starts a token-protected loopback HTTP server. Streaming import/export endpoints. Your local scripts never leave the box.
→ Local REST API
V
Dictate a quick note straight into a folder you pick — your Obsidian vault, your repo’s docs/notes/ tree, your scratch directory. No app stealing focus.
→ Quick Note
VI
Voice triggers map to shell commands behind an explicit allowlist. Run a deploy from the keyboard without touching a mouse. Every run lands in the audit log.
→ Macros

You already drive your editor from the keyboard. Drive dictation from the keyboard too.
Launching April 2026