Frequently asked questions

Minutes. Paste one JavaScript snippet into your site to add the capture widget, or share the hosted feedback URL with no code at all. There is no SDK to wire up and nothing to deploy.

No. The Free Audit runs for 30 days with no credit card. You only pick a paid plan if you decide to continue.

Console logs, network requests, JavaScript errors, a screenshot or screen recording, the page URL and route, and the browser, OS and viewport — attached automatically so engineers can reproduce the issue.

AI writes a title, summary, type and severity for each report, then shows that verdict beside the raw evidence it came from — console errors, failed requests, JS exceptions and on-screen text — so you can verify it.

No. Every AI verdict is shown next to the evidence it was derived from, so engineers can check the work instead of trusting it blindly.

Semantic dedupe uses vector embeddings (pgvector) to merge near-identical reports into a single issue with an occurrence count, rather than relying on exact text matches.

Jira, Linear, GitHub and Slack with two-way status sync, plus Zapier for no-code automation to 6,000+ apps and a signed generic webhook. More providers (ClickUp, Monday.com, Asana, Trello, TestRail, GitLab, Azure DevOps) are in progress.

Yes. When the tracker issue changes status, FeedbackGraph reflects it, and the original reporter is notified automatically when their issue ships — that is the closed loop.

Paid plans run from $8/mo (Starter) through $24/mo (Growth), $59/mo (Pro) and $119/mo (Scale), with custom Enterprise pricing — on top of a free 30-day audit. In India, equivalent INR pricing is billed via Polar (UPI Autopay or card).

Yes. Annual billing saves roughly 20–30% per tier. Subscriptions auto-renew and you can cancel anytime from the billing page.

PII is redacted before the AI sees it. Tenants are isolated with Postgres row-level security enforced on every request. Integration credentials are encrypted with AES-256-GCM. You can export or delete your data yourself for GDPR/CCPA.

Yes. Every team’s data is sealed off at the database with Postgres row-level security, with role-based access control (6 roles) enforced server-side on top.

A read-first Model Context Protocol endpoint that lets your AI agent read FeedbackGraph reports with full evidence, search reports and signals, and read revenue-ranked opportunities — without leaving the agent.

No. It reuses your existing plan’s read paths with scoped, revocable API tokens, so there is no additional cost.

Claude Code, Cursor, VS Code, Windsurf and Slack agents — over stdio (via the mcp-remote bridge) or streamable HTTP.

Yes. One tenant per token, scoped and revocable keys, PII-redacted evidence, per-token rate limits, audit logging, and fail-closed behavior on a bad scope or cross-tenant id. The single write tool (routing) is off by default behind a flag.