Data Use & Privacy Overview

Keystroke does not use your Inputs, Outputs, code, integration data, or run history to train AI models, and we do not permit third parties to train on your data. This applies to all users by default.

The only exceptions, which we describe in the Privacy Policy, are content (1) flagged for security review so we can detect and enforce abuse, (2) explicitly reported to us by you (for example, as Feedback), or (3) content you have explicitly agreed may be used for training.

When you use Keystroke, your prompts and related context (including relevant portions of your project files, tool definitions, integration metadata, and prior conversation history) are sent to third-party AI model providers to generate responses. We use this data solely to generate an Output for the current request and do not use it to train models.

Connection credentials (API keys, OAuth tokens, database passwords, and similar secrets) are never transmitted to AI model providers. Credentials are encrypted at rest and in transit and are used only to call the integrations you've authorized.

Keystroke stores the data we need to provide and improve the Service. That includes, for example:

• Account and organization information, including billing details.
• Your projects, workflows, agents, skills, and the source files that define them.
• Run history and execution traces — including step inputs, outputs, logs, errors, and retries — used for observability, debugging, and audit.
• Integration metadata and encrypted credential material for the third-party services you've connected.
• Usage data such as access timestamps, feature usage, and aggregate performance.

See our Privacy Policy for details on retention.

If you have specific data-handling requirements or questions about how your data is processed, please contact us at team@keystroke.ai.