Run any Skill in Manus with one click

read-whatsapp

Stars2

Forks1

UpdatedJune 12, 2026 at 14:20

Read local macOS WhatsApp chats and summarize user requests, including voice-note transcription. Use when the user asks Codex to inspect a WhatsApp conversation, find what someone said, understand a cancellation/request, search recent WhatsApp messages, or transcribe WhatsApp audio/voice notes from the local WhatsApp database/media folder.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

G33KM44N38

G33KM44N38/dotfiles

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Software DevelopersComputer and Mathematical Occupations·SOC 15-1252

File Explorer

3 files

SKILL.md

readonly

name	read-whatsapp
description	Read local macOS WhatsApp chats and summarize user requests, including voice-note transcription. Use when the user asks Codex to inspect a WhatsApp conversation, find what someone said, understand a cancellation/request, search recent WhatsApp messages, or transcribe WhatsApp audio/voice notes from the local WhatsApp database/media folder.

Read WhatsApp

Use this skill to inspect the user's local WhatsApp Desktop data on macOS. Default to read-only database queries and local transcription.

Workflow

Locate the chat by name, phone/JID, or message text.
Pull recent messages with timestamps, sender direction, text, message type, and media paths.
When audio is present, convert it with ffmpeg and transcribe locally with whisper-cli.
Summarize only the relevant request. State transcript confidence if the model/audio is rough.

Script

Prefer the bundled script:

python3 /Users/boss/.codex/skills/read-whatsapp/scripts/read_whatsapp.py --chat "beautyhairmaiidi" --limit 80 --transcribe

Useful options:

python3 .../read_whatsapp.py --search "annuler" --limit 40
python3 .../read_whatsapp.py --chat "beautyhairmaiidi" --since "2026-06-06" --transcribe
python3 .../read_whatsapp.py --chat "beautyhairmaiidi" --transcribe --download-model

Local Transcription

Use local tools by default; no paid API.

Convert audio: ffmpeg -i input.opus -ar 16000 -ac 1 output.wav
Transcribe: whisper-cli -m ~/.cache/whisper.cpp/ggml-tiny.bin -f output.wav -l fr -otxt

If missing:

brew install whisper-cpp ffmpeg
mkdir -p ~/.cache/whisper.cpp
curl -L --fail -o ~/.cache/whisper.cpp/ggml-tiny.bin \
  https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-tiny.bin

Use larger models when exact wording matters; tiny is fast but approximate.

Data Locations

Default DB:

~/Library/Group Containers/group.net.whatsapp.WhatsApp.shared/ChatStorage.sqlite

Default media root:

~/Library/Group Containers/group.net.whatsapp.WhatsApp.shared/Message

Treat these as private user data. Do not inspect credentials. Do not modify WhatsApp files.

More from this repository

same repository

explain-like-im-five

G33KM44N38/dotfiles

Create a friendly, self-contained webpage that teaches a difficult idea in very simple language using concrete analogies, small visual steps, and playful interaction. Use when the user asks for an ELI5 explanation, says "explain it like I'm five," wants a topic taught visually in a webpage, or needs a beginner-friendly interactive explainer for technical, scientific, financial, or abstract material.

2026-07-102

mailbox

G33KM44N38/dotfiles

Open and operate the user's keyboard-first Apple Mail triage desk with persistent progress and concurrent agent workers. Use when the user says mailbox, wants to review or process email one-by-one, asks to resume inbox triage, wants a local webpage for deciding what to do with messages, or wants agents to handle email instructions while they continue reviewing.

2026-07-102

natural-chat-continuation

G33KM44N38/dotfiles

Use when drafting or sending a reply in an existing chat thread, especially WhatsApp, SMS, email, or DM. Ensures the message continues the current conversation naturally by reading recent context, matching tone, and avoiding unnecessary greetings or restarts.

2026-07-092

herdr-workspace-manager

G33KM44N38/dotfiles

Manage the user's coding workspaces, threads, worktrees, tabs, panes, and running AI agents through Herdr. Use whenever the user says thread, threads, workspace, worktree, tab, pane, agent, switch, focus, open a workspace, create a worktree, start an agent, send input to an agent, check what is running/waiting, or asks to migrate old tmux thread workflows. Prefer this over tmux unless the user explicitly asks for tmux.

2026-07-062

tmux-thread-picker

G33KM44N38/dotfiles

Legacy tmux-only thread picker. Use only when the user explicitly asks for tmux or the old tmux-thread-picker CLI. For normal thread, workspace, worktree, pane, or agent management, prefer the Herdr workspace manager skill.

2026-07-062

send-whatsapp-jules

G33KM44N38/dotfiles

Draft and send WhatsApp messages to Jules from the user's macOS WhatsApp Desktop. Use when the user asks to message, notify, update, or send something to Jules on WhatsApp, especially for BabaCoiffure incident summaries; always draft the message first, apply the user's wording feedback, and send only after explicit confirmation.

2026-06-282

name	read-whatsapp
description	Read local macOS WhatsApp chats and summarize user requests, including voice-note transcription. Use when the user asks Codex to inspect a WhatsApp conversation, find what someone said, understand a cancellation/request, search recent WhatsApp messages, or transcribe WhatsApp audio/voice notes from the local WhatsApp database/media folder.

Read WhatsApp

Use this skill to inspect the user's local WhatsApp Desktop data on macOS. Default to read-only database queries and local transcription.

Workflow

Locate the chat by name, phone/JID, or message text.
Pull recent messages with timestamps, sender direction, text, message type, and media paths.
When audio is present, convert it with ffmpeg and transcribe locally with whisper-cli.
Summarize only the relevant request. State transcript confidence if the model/audio is rough.

Script

Prefer the bundled script:

python3 /Users/boss/.codex/skills/read-whatsapp/scripts/read_whatsapp.py --chat "beautyhairmaiidi" --limit 80 --transcribe

Useful options:

python3 .../read_whatsapp.py --search "annuler" --limit 40
python3 .../read_whatsapp.py --chat "beautyhairmaiidi" --since "2026-06-06" --transcribe
python3 .../read_whatsapp.py --chat "beautyhairmaiidi" --transcribe --download-model

Local Transcription

Use local tools by default; no paid API.

Convert audio: ffmpeg -i input.opus -ar 16000 -ac 1 output.wav
Transcribe: whisper-cli -m ~/.cache/whisper.cpp/ggml-tiny.bin -f output.wav -l fr -otxt

If missing:

brew install whisper-cpp ffmpeg
mkdir -p ~/.cache/whisper.cpp
curl -L --fail -o ~/.cache/whisper.cpp/ggml-tiny.bin \
  https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-tiny.bin

Use larger models when exact wording matters; tiny is fast but approximate.

Data Locations

Default DB:

~/Library/Group Containers/group.net.whatsapp.WhatsApp.shared/ChatStorage.sqlite

Default media root:

~/Library/Group Containers/group.net.whatsapp.WhatsApp.shared/Message

Treat these as private user data. Do not inspect credentials. Do not modify WhatsApp files.