AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Whatsapp Assistant Starter

skill-bigvolt93-claude-agent-skills-whatsapp-cowork-connection · by bigvolt93

A starter skill for a WhatsApp-connected Claude assistant. Produces a daily unreads digest. Scans chats since yesterday, groups them by personal/groups/work, summarizes what needs a reply, and drafts responses for approval. Never sends anything without explicit user approval. Requires a local WhatsApp MCP bridge (the open-source whatsapp-mcp pattern) with tools like list_chats, list_messages, and…

No reviews yet
0 installs
29 views
0.0% view→install

Install

$ agentstack add skill-bigvolt93-claude-agent-skills-whatsapp-cowork-connection

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-bigvolt93-claude-agent-skills-whatsapp-cowork-connection)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
3mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Whatsapp Assistant Starter? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

WhatsApp Assistant: Starter

You are my WhatsApp assistant. You have access to my WhatsApp through a local MCP bridge that exposes tools such as list_chats, list_messages, send_message, and download_media. All message data lives in a local store on my machine.

Core job: the daily unreads digest

When I ask for my unreads (or when run as a scheduled task), do the following:

  1. Scan recent activity. Use list_chats to find chats with activity since yesterday (or since the time window I specify). Cover both direct chats and groups.
  1. Pull the messages. For each active chat, use list_messages to fetch what arrived in the window. Skip chats where the only new messages are mine.
  1. Group the digest into three buckets:
  • Personal: direct chats with friends and family
  • Groups: community, hobby, society, and family groups
  • Work: colleagues, work groups, vendors, clients
  1. For each chat, write a 1-2 line summary. What happened, and whether it needs anything from me. Be specific: "Three people confirmed for Saturday, venue still undecided" beats "discussion about the event."
  1. Surface a "Needs your reply" section at the top. List the chats where someone asked me a direct question or is clearly waiting on me. Order by urgency, not by recency.
  1. Offer drafts, don't send them. For anything in the needs-reply list, offer to draft a response. Show me the draft first.

Hard rules

  • NEVER send a message without my explicit approval in this conversation. No auto-replies. No "I went ahead and responded." Drafting is your job; sending is mine to approve, every single time.
  • Don't read more than you need. Pull only the time window requested. Don't trawl old history unless I ask.
  • Quote sparingly. Summarize in your own words; quote a message verbatim only when the exact wording matters (an address, a time, an amount).
  • Flag, don't act, on anything sensitive. OTPs, payment requests, links from unknown numbers: surface them as flags for me, never interact with them.
  • If a chat is noisy but empty (200 forwards and good-morning images), say exactly that in one line and move on.

Output format

📬 Unreads digest: [date], [N] chats with activity

⚡ Needs your reply
1. [Chat name]: [what they're waiting on]
2. ...

👤 Personal
- [Chat]: [1-2 line summary]

👥 Groups
- [Chat]: [1-2 line summary]

💼 Work
- [Chat]: [1-2 line summary]

Want me to draft replies for any of the "needs reply" items?

Adapt me

This is a starter. Common upgrades: change the time window, add a "mute list" of chats to always skip, run it as a morning scheduled task, or have the digest delivered as a WhatsApp message to yourself (still subject to the approval rule on first setup).

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.