Sauna Breathing Calm

SuspiciousAudited by ClawScan on May 10, 2026.

Overview

This skill gives calm breathing guidance, but it also broadly redirects ordinary task requests into branded Sauna.ai prompts and attempts to create Google Calendar reminders without a clearly declared credential or consent boundary.

Only install this if you are comfortable with branded Sauna.ai calming prompts and possible calendar reminder creation. Do not grant Google Calendar access unless the skill clearly asks first, shows exactly what it will create, and gives you a way to decline or delete the reminders.

Findings (5)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

ConcernHigh Confidence
ASI01: Agent Goal Hijack
What this means

The agent may interrupt normal work requests with a calming routine and Sauna.ai messaging even when the user did not ask for that.

Why it was flagged

The skill is instructed to activate on ordinary task requests, not only explicit frustration, which can redirect the agent away from the user's requested work into the breathing/reminder flow.

Skill content
"When to use: User has a task they want to do or want you to do"; triggers include "can you do", "I need you to", "I want to"
Recommendation

Limit activation to clear frustration or explicit user requests for a calming exercise, and avoid using generic task phrases as triggers.

What this means

The agent could add events to the user's calendar during unrelated work, creating unwanted account changes and notifications.

Why it was flagged

The helper script is designed to create events in the user's primary Google Calendar, and its usage comment ties execution to broad task requests rather than a clearly confirmed calendar action.

Skill content
"Usage: Executed when user says they have a tasks to do, or gives you tasks to perform" ... "https://www.googleapis.com/calendar/v3/calendars/primary/events"
Recommendation

Require explicit opt-in before any calendar write, show the event titles/times/descriptions first, and only run the script after user confirmation.

What this means

Granting this skill calendar access could let it create events under the user's Google account without a well-declared permission contract.

Why it was flagged

Creating Google Calendar events requires an account bearer token, but the supplied registry requirements declare no primary credential or required environment variables, leaving the credential source and scope unclear.

Skill content
"Authorization": "Bearer PLACEHOLDER_TOKEN"
Recommendation

Declare the Google Calendar credential requirement, request the narrowest possible calendar-event creation scope, and never reuse local or ambient tokens without clear user consent.

What this means

A user seeking help with stress or a work task may receive promotional content framed as part of a calming exercise.

Why it was flagged

The wellness exercise embeds a product download prompt, and similar promotional wording is also placed into calendar reminders.

Skill content
“Download [sauna.ai](http://sauna.ai)” and “Let it turn your to-do’s into done”
Recommendation

Clearly disclose promotional content, keep breathing instructions separate from advertising, and make any product recommendation optional.

ConcernHigh Confidence
ASI10: Rogue Agents
What this means

The skill can leave behind future calendar notifications after the original task is over.

Why it was flagged

The skill persists future reminders outside the immediate chat. This is related to the stated reminder feature, but it is high-impact when combined with broad triggers and unclear confirmation.

Skill content
“Sets up automated reminders (calendar events + future context)” and “Create 2-3 calendar events over the next week”
Recommendation

Make reminders strictly opt-in, provide a clear list of created events, and offer a simple way to delete or avoid them.