Privacy
We built Stars Engine because we believe tools that map the human soul should not be built by ad-tech executives. This isn't a privacy policy written by lawyers to protect us. It's a promise to you.
The Short Version
✓ We don't sell your data. We don't share it. We don't analyze it for anyone but you.
✓ We don't have advertisers. We don't have trackers. We don't have analytics scripts.
✓ We don't store your quiz answers on our servers. You take a quiz, the engine scores it, the raw responses are deleted immediately. Only the calibration result is kept. Optionally, you can choose to save your raw answers to your own device — they stay under your control.
✓ Chart computation is deterministic and runs entirely on our own hardware. When you ask for an AI-written narrative, we send the chart interpretation to an external model with your name, birth date, birth place, and coordinates stripped out first — never your raw birth data.
✓ You can export everything. You can delete everything. One click, gone.
✓ We run on our own servers. No AWS. No Google Cloud. No third-party infrastructure that could be subpoenaed without our knowledge.
What We Store
Birth data
Date, time, and location of birth. This is required to compute a chart. Stored in PostgreSQL on a dedicated machine in Texas. Encrypted at rest.
Email address
Used for account login and password recovery. Never shared. Never added to any mailing list without explicit opt-in.
LLM preferences
Tone, complexity, and focus area settings for narrative generation. Stored as JSON. You control these.
Calibration results
When you take a psychometric quiz, the raw item responses are processed immediately, scored against our engine's predictions, and then deleted from our servers. Only the resulting calibration profile — the comparison between your chart's prediction and your self-report — is retained. We do not store the individual answers you gave.
Chart archive
Every computed chart is archived so results stay reproducible and we can prove a reading was not altered after the fact. The archive is stored on our own hardware in Texas and is organised by an identifier derived from the name and birth date you entered. It is not anonymous — anyone with access to that machine could tell which chart belongs to whom. Access is limited to the operators of this service. Deleting your account removes your charts from the archive.
Local storage (optional)
After completing a quiz, you have the option to save your raw answers to your own device (browser localStorage). These answers never touch our servers — they live on your machine. If you're a therapist who wants to keep a record, or if you want to allow richer LLM-calibrated narratives that reference your specific answers, you can opt in. By default, answers are ephemeral — processed and deleted.
What We Never Store
Quiz item responses. The 293 Likert-scale answers you give across 12 batteries exist only in memory during scoring. Once the calibration result is computed, the raw answers are purged. We keep the outcome, not the journey.
LLM prompts or completions. When you generate a narrative, the prompt is assembled, sent to the LLM provider with PII scrubbed, the response is returned to you, and neither the prompt nor the completion is stored. We don't train on your data. We don't keep a history of your conversations with the engine.
Client data you don't own. If you're a practitioner, your clients' charts belong to you. We don't email them. We don't market to them. We don't even know who they are beyond the birth data you entered for computation.
Behavioral analytics. No Google Analytics. No Mixpanel. No Hotjar. No session recording. We don't know which pages you visit or how long you spend on them.
Infrastructure & Security
Stars Engine runs on self-hosted bare-metal servers. The compute engine in Texas handles all chart calculation and database operations. The web frontend runs on a VPS in Germany. All traffic between them travels over an encrypted WireGuard tunnel.
Passwords are hashed with scrypt. Authentication uses short-lived JWT tokens (72-hour expiry). API keys use SHA-256 hashing. The MCP server requires bearer token authentication.
We use a single httpOnly cookie for session management. No tracking cookies. No third-party cookies. No fingerprinting.
Your Rights
Export. Download all your data — chart data as JSON, documents as Markdown, calibration results. One click from Account Settings.
Delete. Permanently erase your account and all associated data. Charts, documents, quiz results — everything. Gone from our servers within 24 hours.
Opt out of research. The research dataset is opt-in only. Your chart data is never included in population statistics or convergence studies without your explicit consent. The default is private.
No data portability trap. We use open formats (JSON, Markdown, Parquet). Your data is yours. Take it anywhere.
Breach Notification
If we discover a data breach affecting your personal information, we will notify affected users within 72 hours of confirmation. Notifications will be sent to the email address associated with your account. We maintain access logs and intrusion detection on all production servers.
What We Won't Do
We will never sell your data — not to advertisers, not to data brokers, not to research institutions. We will never run ads. We will never use your chart data to train AI models without your explicit consent. We will never share your email address. We will never contact your clients.
If we receive a legal request for your data, we will notify you before complying, unless legally prohibited from doing so. We have never received such a request.
This policy was last updated July 2026. It will never get longer or more complicated than this. If we change how we handle your data, we'll tell you directly — not bury it in a 40-page document you'll never read.