Security audit
Security checks for vulnerabilities and agentic risk
This skill is a coherent Habit AI API guide, but users should understand it sends sensitive health, journal, and photo data to an external service.
Install only if you intend to use Habit AI as a cloud service for personal health tracking. Treat journal entries, profile details, food photos, and coaching messages as sensitive data sent to Habit AI, keep the API key private, and ask the agent to confirm before deleting records, changing profile data, uploading images, or managing API keys.
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
Update meal fields: `mealName`, `calories`, `protein`, `carbs`, `fat`, `fiber`, `sodium`, `sugar`, `ingredients`, `mealType`, `healthScore`, `imageUrl`, `analysisConfidenceLevel`.
### DELETE /meals/:id
Delete a meal.
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
Upserts per day. Also updates profile `weightInKg`.
### DELETE /weight/:id
## Meditation
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
Required: `durationInMinutes` (number)
Optional: `type`, `notes`
### DELETE /meditation/:id
## Journal
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
Required: `content` (string)
Optional: `topic`, `mood`
### DELETE /journal/:id
## Saved Meals
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
Required: `name` (string)
Optional: `calories`, `protein`, `carbs`, `fat`, `fiber`, `sodium`, `sugar`, `ingredients`, `mealType`, `imageUrl`
### DELETE /saved-meals/:id
## AI Analysis
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
List active keys (prefix only, not full key).
### DELETE /keys/:id
Revoke an API key.
The skill explicitly instructs sending highly sensitive personal data to a third-party API, including health, nutrition, profile, meditation, and journal information, but does not warn the user that this data leaves the local environment or advise obtaining explicit consent. Because the data categories are unusually sensitive, the omission increases privacy and compliance risk and could cause users to disclose intimate information without informed awareness.
The photo-upload workflow sends user meal images to an external service and returns a cloud storage URL, but the skill provides no warning that images are uploaded off-device and may become retrievable via a remote link. Food photos can contain faces, location clues, medical items, or other sensitive context, so failing to disclose this external storage behavior materially raises privacy risk.
This command transmits base64-encoded user image data to an external API, which is expected functionality for the feature but still constitutes a real external data exfiltration path. In this skill's context the transmission is not inherently malicious, yet it is dangerous if performed without clear user awareness because the payload may contain sensitive visual information and is tied to an authenticated account.
If you have a food photo, upload it first to get a URL:
curl -X POST https://habitapp.ai/api/v1/meals/upload-image \
-H "Authorization: Bearer $HABITAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"imageBase64": "<base64-encoded-image>"}'
The food-image analysis endpoint accepts raw base64 image uploads, which may contain sensitive personal content beyond food, including faces, location clues, or metadata-derived context. Without an explicit warning in the skill, users may unknowingly upload private images to remote AI analysis.
The coach endpoints explicitly state they use the user's meal history, goals, patterns, and meditation history, which can include sensitive health and behavioral data. Without a clear privacy warning or consent boundary in the skill context, an agent could transmit intimate user data to AI processing in ways the user does not fully expect.
The manifest describes using Habit AI for logging health data, journaling, coaching, and food analysis, but this reference also exposes creation, listing, and revocation of API keys. API credential management is a separate account-administration capability not mentioned in the skill description and materially broadens what the documented API can do.
No suspicious patterns detected.