agentleFS
Sign inStart free
kaithoughtarchitectkaithoughtarchitect/toppingout/skills/speak/SKILL.md

speak

Skill14 starsChanged 20 days agoMITView on GitHub

Explain my last response aloud, rewrite it into a simple spoken explainer and play it via Grok Voice TTS. Use when the user runs /speak, says "say that out loud", "read that back", "explain that aloud", or when an output is dense and they want to listen instead of read. One reply at a time: to keep it on for every reply in the session, use converse.

  • Reads credentials

18 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 17 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run every command from the repo root.

    line 28
  2. Very specific:

    Source text: if arguments were passed to /speak, explain that text.

    line 34
  3. Distinctive:

    No code, no file paths, no markdown, no headings, no bullet points read aloud, just natural spoken sentences.

    line 41
  4. Distinctive:

    Keep it tight: ~20–60 seconds of speech (roughly 60–180 words).

    line 42
  5. Distinctive:

    If the source was already short and simple, just smooth it for speech rather than padding it.

    line 43
  6. Very specific:

    Write the rewritten explainer to scratch/speak-input.txt in the repo root (overwrite it.

    line 47

“Common” is measured: kept by at least 6 in 10 of the 17 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Speak

Turn my last message into a short spoken explainer and play it as audio. For when an output is dense or complex and I want to listen to a plain-language version of it.

Usage

/speak            # explain my previous message aloud
/speak <text>     # explain/voice the given text instead

Plays through the speakers. Default engine is Grok Voice TTS via OpenRouter (voice leo). Configure it in the repo's .env (see "Voice setup" below). Needs Node with npx tsx, the bundled scripts/speak.ts + scripts/bootstrap-env.ts in the repo, and Windows audio playback. Run every command from the repo root; scratch/ is a gitignored working folder there.

Instructions

Source text: if arguments were passed to /speak, explain that text. Otherwise use your immediately preceding assistant message (the one right before this /speak invocation).

Rewrite it, do not read it verbatim:

  • Plain, conversational tone, like a sharp friend explaining it to me over coffee. Warm, casual, not verbose.
  • Explain the gist and the why in simple terms. Drop jargon, or define it in a few words if it actually matters.
  • No code, no file paths, no markdown, no headings, no bullet points read aloud, just natural spoken sentences.
  • Keep it tight: ~20–60 seconds of speech (roughly 60–180 words). Hit the key point(s), not every detail. Hard cap ~4000 characters.
  • If the source was already short and simple, just smooth it for speech rather than padding it.

Then:

  1. Write the rewritten explainer to scratch/speak-input.txt in the repo root (overwrite it; create scratch/ if missing).
  2. Run: npx tsx scripts/speak.ts scratch/speak-input.txt

Reply with only 🔊 (plus a one-line note if the script printed an error). Do not show the rewritten explainer text unless I ask for it.

Voice setup / changing the narrator

/speak picks its engine from the repo-root .env (loaded by scripts/bootstrap-env.ts) and uses only that one (no fallback).

  • SPEAK_PROVIDER, grok (default) or hume.
  • Grok (default), xAI Grok Voice TTS via OpenRouter (OPENROUTER_API_KEY). ~10× cheaper than Hume, no setup step. Pick the voice with SPEAK_GROK_VOICE, one of eve (energetic), ara (warm), rex (confident), sal (balanced), or leo (authoritative, the default). These 5 are the only voices OpenRouter exposes; custom/cloned voices need xAI's native API.
  • Hume, the saved Octave custom voice (SPEAK_VOICE_ID). Created once by scripts/speak-setup.ts, which designs + saves a Hume voice and prints a SPEAK_VOICE_ID for .env. Saved Hume voices are permanent (an id always resolves to the same voice), so switching/reverting is editing that one line. To mint a new one: edit VOICE_DESCRIPTION in that script, run npx tsx scripts/speak-setup.ts, then update SPEAK_VOICE_ID.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free