Privacy in the current preview

Updated October 3, 2026.

Your reports

BugFiled receives the report fields sent by ChatGPT or Codex to prepare drafts. Its application does not save report bodies or screenshots to a database. Report content is processed in memory. Duplicate searches and Jira field lookups send the required query or project details to your configured tracker. Issue submission and attachment uploads are not implemented.

Optional usage metrics

Metrics are off for each tool call unless you explicitly opt in. If enabled, the application logs the tool name, Jira/GitHub provider, success or failure, execution duration, counts of filled fields and open questions, whether screenshot evidence is present, duplicate candidate count, and Jira metadata field count. These are simple usage and reliability signals; they do not establish report accuracy.

Metrics exclude report text, prompts, images, filenames, repository and project names, URLs, user identifiers, conversation identifiers, IP addresses, and raw error messages.

Your choice

Tell BugFiled “Do not track usage metrics” to keep metrics off, or “I agree to content-free usage metrics” to opt in for this conversation. You can change your choice at any time. Clients communicate the choice with the optional metricsConsent tool argument. Leaving it out is an opt-out. This controls BugFiled metrics, not logging by ChatGPT, Codex, Render, or your tracker.

Storage and retention

Opted-in metric events are written to the server's standard output. The hosted service uses Render, whose service logs attach timestamps and retain logs according to the service's plan and settings. Local installations write to the local server console. BugFiled has no separate analytics database or PostHog connection. Hosting providers may separately process network information for operating their service.

This notice describes the current prototype. Any future account connections, persistent report storage, or analytics services will require an updated notice.