Dictate in code mode
Detects IDE and terminal apps and leaves identifiers, quotes, and bracketing alone. Dictation tokens (“new line”, “in parens”) become characters — everything else stays literal.
Skald for developers
Typing is the bottleneck, not the thinking.
The code in your editor is clear in your head. The commit message, the PR comment, the Slack reply to the reviewer — that's what stands between you and the next task. Skald takes over that writing without making you switch to a text editor.
A typical day
A morning with Skald inside the IDE — three places where Skald fills the typing pause.
- 01
Commit message in the terminal
You're in the shell, changes staged. Hold the hotkey: “Fix race in token-refresh — token is now hydrated once per request, not per method call.” Release. In code mode, Skald recognises identifiers and path tokens and leaves them alone — no word fixes, no inserted semicolons.
- 02
PR-review comment in the browser
GitHub PR open, cursor in the review-comment box. You speak what you noticed: “Use the hash as input rather than the raw token — otherwise the first refresh leaks the cleartext into the logs.” Skald reads the window title, recognises a PR-review context, keeps it tight, holds your tone.
- 03
Brain-Dump for an architecture sketch
You open the Brain-Dump window and talk for ten minutes about the next iteration — trade-offs, open questions, risks. Skald returns structured Markdown: TL;DR, sections, action items. You paste into Linear, fix two words, done.
The modes you'll use most
Command for PR diffs
Select, hotkey, speak: “shorter”, “in English”, “as a list”. The selection is replaced; the rest of the comment stays.
Brain-Dump for ADRs
Long thoughts in, structured Markdown out. TL;DR on top, sections, action items at the bottom — paste-ready for Linear or Notion.
Where Skald shows up
Try Skald
Skald is in closed beta. Join the waitlist — we'll get in touch when your role is up next.