Local vs cloud
Edit, build, flash, debug, and serial stay on your machine.
When you use HardwarePilot, the IDE sends a bounded prompt context to Prescaler's cloud proxy (app.prescaler.ai), which forwards it to the model provider (Anthropic).
That context can include open files or selection, SVD snippets, serial lines, build errors, debug stack or locals, Agent-read file contents, and datasheet text excerpts when grounding is on.
Do not assume source never leaves your laptop when live AI is enabled.
Uploaded datasheets
Datasheet PDFs are imported and parsed locally in the IDE.
Prescaler does not treat the whole PDF as a cloud upload for AI by default.
When HardwarePilot uses datasheet grounding, extracted text excerpts may be included in prompts sent through the proxy.
Accounts and retention
Sign-in uses GitHub, Google, or email through Prescaler's account service.
Usage metering for Free request counts and paid plan dollar credits is stored with your account so quotas can be enforced.
Prompt content is processed to fulfill the AI request. For retention and subprocessors, see the Privacy Policy.
What we do not offer yet
Offline or air-gapped inference is not available.
Enterprise SSO is not supported in the product today.
Security questions
Email [email protected] for security questions. For product bugs or IDE issues, use GitHub Issues.