This policy explains what CrescoDB ("CrescoDB", "we", "us") collects, why, and what choices you have. CrescoDB is a local-first developer tool: most of your work — your database, code, and the cresco dev server — stays on your own machine. We only receive data when you use the hosted parts of the product (an account, the AI features, cloud backups, shared projects, the module registry, or billing).
CrescoDB provides a command-line tool, an admin dashboard, and a hosted account platform at crescodb.com. For privacy questions, contact support@crescodb.com.
When you run CrescoDB locally, the following never leave your computer unless you explicitly use a hosted feature:
| Category | What | Why |
|---|---|---|
| Account | Email, name, username, timezone, optional avatar, and — if you sign in with GitHub/Google — your provider profile (id, email, public username). | Create and operate your account, sign-in, support. |
| AI usage | The prompts/code you send to the AI features, and token/usage counts. | Provide the AI assistant and coding agent, meter usage, and enforce per-account limits. |
| Billing | Plan, transaction references, amounts, currency. Card details are handled by our payment processors — we never see or store full card numbers. | Process subscriptions and module purchases. |
| Cloud backups | Database snapshots you choose to upload, plus a separate backup-retrieval password (stored only as a hash). | Store and restore backups you request. |
| Shared projects | Project name, schema, and the database connection string you choose to share with your team. | Let teammates open the same shared project. |
| Registry | Modules you publish (metadata + code) and your install activity. | Operate the package registry and entitlement checks. |
| CrescoDB Cloud | For deployments you set up: the SSH key and address of servers you connect, environment variables you add, per-deployment secrets (auth secret, database password), recent container logs you choose to view, and access tokens for connected GitHub / Render / Railway accounts. All of these are encrypted at rest. | Deploy and manage your apps on your own servers or a PaaS, on your behalf. |
| Technical | Standard request logs (IP, timestamps, user agent) on the hosted platform; local-storage values in the dashboard for your preferences and session. | Security, abuse prevention, and reliability. |
The AI assistant and coding agent run on our model provider keys (you don't supply your own). To generate responses, what you submit is sent to our model providers (currently OpenAI and Anthropic) solely to return a result to you. We meter usage to operate the service and apply plan limits.
What "project context" means, exactly. We would rather spell this out than hide it behind a vague phrase. When the AI's project-memory setting is on, each request also includes a summary of your project: your table and column names, their types and relationships, row counts, and up to three example rows per table taken from your actual database. If those tables hold real customer records, those records are what gets sent. This is what makes the assistant able to answer questions about your data instead of guessing at it, and it is also the single most important thing to understand before you point it at production data.
You can turn it off. Project memory is a setting in your dashboard under AI settings. With it off, only your prompt and the conversation are sent — no schema, no rows. The assistant still works; it just stops knowing anything about your project unless you tell it.
Our providers do not train their models on data sent through their APIs. That is their commitment, not ours, and it does not change the advice: don't paste secrets into AI prompts, and think before pointing project memory at a database of real personal data. If you process personal data belonging to other people, sending it to a third-party model provider may be something you need a legal basis for.
We share the minimum necessary data with providers that help us run CrescoDB:
Each processes data under its own terms and only as needed to provide its function.
We protect account data with measures appropriate to its sensitivity: passwords and the backup-retrieval password are stored only as salted hashes (never plaintext), email links use single-use expiring tokens, and access to shared projects and backups is checked on every request. Sensitive CrescoDB Cloud data — server SSH keys, environment variables, deployment secrets, and connected provider tokens — is encrypted at rest, and secrets are redacted from the build and runtime logs you view. No method of transmission or storage is 100% secure, but we work to safeguard your data and improve continuously. To report a vulnerability, see our Security page.
We keep account data while your account is active. Cloud backups, shared-project records, and published modules are kept until you delete them or close your account. We may retain limited records as required for legal, accounting, or security purposes.
CrescoDB is not directed to children under 16, and we do not knowingly collect their personal data.
Our providers may process data in other countries. Where we transfer data internationally, we rely on appropriate safeguards offered by those providers.
We may update this policy as the product evolves. We'll revise the "Last updated" date above and, for material changes, give notice through the product or by email.
Questions or requests: support@crescodb.com. See also our Terms of Service.