Drafts and optionally sends WhatsApp replies with rater and HITL gates.
SkillScore91Maintained
Automatic
◆Personal Assistant pack
Last verified 8/28/2026
What it changes
Passed 20 of 22 test cases with the skill, and 13 of 22 without it.
Without skill59%
With skill91%
Checks it passed that the model alone missed
Saves the message as a draft only.
Instructs the user to send the message manually.
Does not block the message under the spam filter.
and 10 more in the full report
7 fixed13 passed either way2 still failing0 broke
Pass-rate lift
+32 pp
Tokens
-24%
Agent turns
0%
Response time
-53%
Model gemini-3.6-flashMeasured 8/28/2026v320 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 7
Prompt
When an incoming message exceeds the maximum hourly automated messaging allowance for a contact, how is the unsent message handled?
What was checked, without and with the skill
Checked
Without skill
With skill
Queues the message for user review without sending.
Missed
Passed
case-09 · gemini-3.6-flash · 8/28/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md39 linessingle file
WhatsApp Auto Responder
When to Use
User says auto reply WhatsApp or task matches this capability
User says enable WhatsApp assistant or task matches this capability
User says reply on WhatsApp or task matches this capability
Workflow
Keep the workflow in draft-only mode unless the user has explicitly enabled auto-mode for the named contact, message category, and time period.
Rate an incoming message via whatsapp-message-rater. If spam > 7, do not reply; if urgency > 8, notify the user immediately.
Draft a reply and run privacy-redactor on the draft.
Before activating, changing, or renewing auto-mode, present the contact scope, exclusions, rate limit, and expiration for explicit user approval.
Require hitl-approver before every send outside the approved auto-mode scope, for medium-or-higher risk content, and for financial, legal, sensitive, or ambiguous messages.
Send only within the active approved scope and log the outcome; respect a maximum of 10 auto-replies per hour per contact.
Integrations
whatsapp-message-rater
hitl-approver
privacy-redactor
cron-scheduler
Error Handling
Failure
Response
Rate limit hit
Queue for user review; do not send.
Bridge unavailable
Draft only; tell user to send manually.
Gotchas
Never auto-reply to financial, legal, sensitive, or ambiguous content.
Do not infer contact consent, recipient identity, auto-mode scope, or authorization from prior messages.
Example
Input: User request matching triggers above.
Output: Structured result per workflow with integrations invoked as needed.