agentleFS
Sign inStart free
tylerthebuildortylerthebuildor/yap/skill/SKILL.md

yap

Skill0 starsChanged 58 days agoMITView on GitHub

Control "yap", the global text-to-speech that reads Claude's finished message aloud on macOS. Use whenever the user wants to turn spoken output, voice output, read-aloud, or TTS on or off, silence it right now, change the voice, speed or length cap, check whether it is on, or debug why it did or did not speak. Triggers on the word "yap" and on phrases like "stop reading your answers out loud" or "turn the speech back on".

4 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions4 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 4

    1. dzhng/yap★ 112same name
    2. duncan-buildroom/profit-room-skills★ 18same name
    3. Wondermonger-daydreaming/claude-skills-library★ 7same name
    4. karloows/orchraft★ 1same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 4

    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. Distinctive:

    Run the CLI. Do not edit the state files by hand.

    line 11
  2. Very specific:

    If yap is not on PATH in the current shell, call ~/.local/bin/yap directly.

    line 24
  3. Very specific:

    Never repeats: the uuid of the last spoken message is stored per session in ~/.claude/yap/state/.

    line 56
  4. Very specific:

    Never stale: a message older than FRESH_WINDOW is skipped rather than spoken late.

    line 57
  5. Distinctive:

    Never overlaps: one global lock, so two sessions finishing together queue instead of talking over each other.

    line 58
  6. Distinctive:

    Never blocks: the hook is async and the speech is detached from the session.

    line 59
  7. Distinctive:

    Never breaks a turn: every path exits 0.

    line 60

“Common” is measured: kept by at least 6 in 10 of the 4 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

yap

yap speaks Claude Code's final message of each turn through macOS say. It is a global Stop hook, so it applies to every Claude session on this machine.

Run the CLI. Do not edit the state files by hand.

yap on          # start speaking finished messages (all sessions)
yap off         # stop, and silence anything speaking right now
yap toggle
yap status      # is it on?
yap stop        # shut up right now, but stay on for the next message
yap test [text] # speak a test line
yap log [n]     # what it did or skipped recently
yap doctor      # wiring, dependencies, active speech

If yap is not on PATH in the current shell, call ~/.local/bin/yap directly.

Answering "is it working?"

yap doctor shows everything: the on/off flag, whether the Stop hook is registered in ~/.claude/settings.json, and whether jq, say and perl are present. yap log 20 shows recent decisions, including why a message was skipped.

Changing behavior

Settings live in ~/.claude/yap/config (create it if absent, it is sourced by the script). Defaults are at the top of the script itself; yap doctor prints its path.

VOICE="Samantha"    # say -v '?' lists installed voices; empty = system default
RATE=190            # words per minute; empty = system default
MAX_CHARS=1400      # longer messages are cut at a sentence boundary
SPEAK_CODE=0        # 1 = read fenced code blocks aloud instead of saying "code block"
FRESH_WINDOW=25     # never speak a message older than this many seconds

Config changes apply to the next message in every session, with no restart.

How it decides what to speak

Reads the session transcript named by the hook payload, takes the last assistant entry that contains real text, and speaks only that. It skips tool calls, subagent output, and the user's own words. Markdown is flattened, URLs become the word "link", and fenced code becomes "code block".

Guards, in case something looks wrong:

  • Never repeats: the uuid of the last spoken message is stored per session in ~/.claude/yap/state/.
  • Never stale: a message older than FRESH_WINDOW is skipped rather than spoken late.
  • Never overlaps: one global lock, so two sessions finishing together queue instead of talking over each other.
  • Never blocks: the hook is async and the speech is detached from the session.
  • Never breaks a turn: every path exits 0. The worst failure is silence.

Where it lives

The script is a git checkout, symlinked into place, so editing the checkout changes the live tool with no reinstall. yap doctor prints the real paths:

  • the script: a clone of https://github.com/tylerthebuildor/yap
  • the command: ~/.local/bin/yap, a symlink to that script
  • the skill: ~/.claude/skills/yap, a symlink to skill/ in the same checkout
  • state, log and config: ~/.claude/yap/, outside the checkout

To update it: git pull in the checkout. To change install wiring: make install.

Scope

Turning it off is global and immediate, including for sessions that are already running. Turning the hook itself on or off (as opposed to the flag) means editing hooks.Stop in ~/.claude/settings.json, which only affects newly started sessions.

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