Send Messages by Voice.
Never Open Slack.

Step 1: Create a Slack Incoming Webhook

Each webhook is tied to one channel. Create one per channel you want to post to.

1. Go to https://api.slack.com/apps → Create New App → From scratch
2. Name it "Voice Messages", select your workspace
3. Click Features → Incoming Webhooks → toggle Activate Incoming Webhooks to On
4. Click Add New Webhook to Workspace → select the channel → Allow
5. Copy the webhook URL: https://hooks.slack.com/services/T.../B.../xxx

The webhook URL is the only credential you need. Pass it as SLACK_WEBHOOK_URL inline in the exec line — no API key required.

Step 2: Add integrations/send_slack.py

A small Python script that reads the transformed text from stdin and posts it to Slack via webhook. No external dependencies — stdlib only.

#!/usr/bin/env python3
"""Send a message to Slack via Incoming Webhook."""
import json, os, sys, urllib.request

def main():
    text = sys.stdin.read().strip()
    if not text:
        sys.exit(0)

    url = os.environ.get("SLACK_WEBHOOK_URL")
    if not url:
        print("SLACK_WEBHOOK_URL not set", file=sys.stderr)
        sys.exit(1)

    req = urllib.request.Request(url,
        json.dumps({"text": text}).encode(),
        {"Content-Type": "application/json"})
    resp = urllib.request.urlopen(req, timeout=10)
    print(f"Slack ({resp.status}): {text[:60]}...")

if __name__ == "__main__":
    main()

Step 3: SpeechButton config.toml

One hotkey per channel. The transform field points to a prompt file — SpeechButton runs the local AI model against it. The exec field sends the result to Slack.

# ~/.config/speechbutton/config.toml

[global]
model = "parakeet-tdt-0.6b-v3-int8"

[parakeet-tdt-0.6b-v3-int8]
language = "en"

# Slack — local AI transform (recommended)
[[hotkey]]
key = "RightCommand"
channel = "5"
name = "slack"
transform = "prompts/slack_message.md"
exec = "SLACK_WEBHOOK_URL=https://hooks.slack.com/services/T.../B.../xxx integrations/send_slack.py"

transform = "prompts/slack_message.md"

transform = "transforms/transform_claude.py prompts/slack_message.md"

Start with Option A — it's free and works offline. Switch to Option B only if you need higher quality.

Step 4: Prompt files — tone per channel

Each channel gets its own prompt file in the prompts/ folder. Same voice, different audience, different output.

Clean up this spoken message for a Slack dev channel.
Keep it casual and concise. Fix grammar, remove filler
words (um, uh, like, basically), keep technical terms.
Output ONLY the cleaned message, no quotes.

Rewrite this spoken message for a professional Slack
channel. Make it clear, well-structured, and polished.
Use bullet points if there are multiple items.
Output ONLY the message.

Format this spoken standup update into three sections:
*Yesterday*, *Today*, *Blockers*. Use bullet points.
If no blockers mentioned, write "None".
Output ONLY the formatted standup.

Prerequisites

• macOS 15+ (Sequoia), Apple Silicon
• A Slack workspace where you can create apps
• Python 3 (pre-installed on macOS)

Quick Start

1. 1 Download SpeechButton — free 15 minutes/day, no account needed
2. 2 Create a Slack webhook at api.slack.com/apps — Features → Incoming Webhooks → Add New Webhook to Workspace
3. 3 Save send_slack.py to ~/.config/speechbutton/integrations/
4. 4 Write a prompt file in ~/.config/speechbutton/prompts/slack_message.md with the tone instructions for your channel
5. 5 Add a [[hotkey]] block to config.toml with your webhook URL in the exec line
6. 6 Hold RightCommand, speak, release. Your first voice message arrives in Slack in under 5 minutes.