CALVRAS
KNOW WHAT YOU SHARE

Privacy Policy

How Calvras handles project context, local storage, and external AI requests.

Updated · 14 October 2026
01

Operator and scope

Calvras operates the website and desktop coding agent. Contact calvrasnoreply@gmail.com with privacy questions or requests. This policy explains the current local storage and external-service flows. A public street address and country are not supplied here; no storage country or jurisdiction-specific certification is claimed.

02

Project and conversation information

Tasks can involve prompts, conversation history, selected source files, attachments, project paths, and tool output. That information can contain personal or confidential data. Only provide information you are authorized to share, and remove unnecessary identifiers or secrets before submitting a task.

03

Desktop storage

The native desktop code uses a local SQLite database under the Calvras application-data folder for project references, chat sessions, messages, and settings. Local project files remain in their project folders. Removing a project reference is different from deleting the underlying files or all stored conversation records.

04

Browser storage

The web workspace stores project references, recent activity, and the active project in browser local storage. User-provided Gemini and OpenRouter keys may also be stored there. Clearing site data removes browser-stored values but does not remove desktop databases, local files, backups, or information already sent to external services.

05

Credential handling

The native Windows credential service stores secrets locally using Windows protection tied to the current user. The web workspace’s browser storage is a separate mechanism and should not be treated as a secure credential vault. Protect access to your device and profile, and rotate keys through the provider if exposed.

06

AI request processing

Depending on configuration, prompts and selected context are sent through the Calvras API or directly to a model provider. Current integrations include Google Gemini, OpenRouter, and Hugging Face. Providers process the request to generate a response. Local file execution does not mean that model processing is local.

07

Purposes of processing

Prompts and context support requested AI responses and coding tasks. Local project references, messages, and settings support workspace continuity. Command output helps the agent assess a result and decide its next action. Information supplied in a support or partnership message is used to understand and respond to that inquiry.

08

Hosting, downloads, and providers

Cloudflare hosts the website and GitHub delivers the linked installer and portable download. AI services and email providers have their own processing practices. When you follow a provider or download link, that destination may process connection information under its own terms and privacy policy.

09

Logs and service metadata

Hosting, model, and email services may process request metadata and operational logs needed to deliver or protect their services. Those practices depend on the service and account configuration. This policy does not claim that every provider has logging disabled or that all requests are immediately deleted.

10

Retention and deletion

Locally stored workspace and conversation information persists until it is removed or the relevant local data is cleared. External-provider request and log retention varies by provider and configuration. No universal retention period is asserted. Contact Calvras about information it controls; separate provider records may require a request to that provider.

11

Support and partnership messages

If you email Calvras, the message, email address, and attachments pass through email services and reach the recipient. Include only the information needed for the inquiry. Ask for deletion of inquiry information by emailing the contact address; applicable obligations and the systems involved may affect what can be removed.

12

Your choices

Choose the project folder and context you share, review tool actions, and avoid providing unrelated personal information. Clear browser site data when needed, manage desktop data separately, and revoke credentials through their providers. These actions cannot guarantee removal of a copy already processed elsewhere.

13

Privacy rights and requests

Depending on applicable law, you may have rights to access, correct, delete, restrict, object to processing, or obtain a portable copy of personal information. Email calvrasnoreply@gmail.com with enough non-sensitive detail to identify the relevant interaction. Applicable identity checks and legal exceptions may affect a request; do not send passwords or API keys.

14

International processing

Hosting, model, download, and email services may process information outside your location. Consult the provider’s current terms and privacy documentation for region and transfer information. This policy does not claim a specific processing country or an audited transfer arrangement for every connected service.

15

Security and incident reports

Keep your device and software current, restrict access to project folders, and protect credentials. No system can be assumed risk-free. Report suspected privacy or security issues to calvrasnoreply@gmail.com with redacted details. Do not include live secrets, a working exploit, or someone else’s private data in an initial email.

16

Cookies and policy changes

The current workspace uses browser local storage; this is distinct from cookies. Provider and infrastructure practices may change, so this policy does not promise that no tracking or logging exists anywhere in the service chain. New data uses should be reflected in the published policy and update date. Review it before sharing sensitive information.

Privacy disclosures should reflect actual practices. Read the FTC’s privacy guidance.

Explore the plans