Feedbacks: UI Context for AI Agents

Capture website feedback, screenshots and session replay for AI coding agents. Connect to your team's Feedbacks server.

Turn website feedback into context your team and AI coding agents can act on. Feedbacks connects a Chrome review to your team's Feedbacks server. Clients and teammates can point to a specific element, comment on a screenshot, or capture a complete page. Developers can inspect the same page context, discussion and approved evidence before making a change. CAPTURE AND REVIEW • Right-click a point on a page to comment, or open the compact popup to capture the current view. • See the current capture stage on the page while screenshots and page context are prepared; the status stays out of saved images. • Keep several points in a local draft, then review the screenshots before sending. • Mark screenshots with pencil, shapes, arrows and text; redact visible private information before sharing. • Capture a full page when one view is not enough. Open page comments and resolved points when you need the team's history. • Keep the on-page review controls in view while opening more tools and options. VIDEO AND SESSION CONTEXT • Record a short tab video with a synchronized browser session, or choose session-only capture in the recording options. • Review video or page replay with activity, console, network and other available events on one synchronized timeline. Use a larger player or place the event inspector beside it on a wide screen. • Select an event to inspect that moment, and save or annotate a video frame for the thread. Coverage labels show when a channel or part of the session was unavailable. • Network request and response bodies are a separate recording option. Page text and input masking can be enabled. Video pixels need their own visual review before sending. FOR DEVELOPERS Feedbacks keeps submitted comments, screenshots, page URLs and recording evidence in the project on your chosen server. Authorized teammates can download a complete thread bundle with its discussion, page context, approved images, recordings and available diagnostics for offline review. Developers can copy a focused task link for a coding agent. With permission, the agent can start with the problem and one relevant screenshot or saved video frame, then request more recording or diagnostic evidence only when needed through Feedbacks MCP or API. Connecting an agent does not automatically make changes or send data to an AI provider. SETUP Your organization needs a Feedbacks server and an account first. Self-host the open-source application or ask your team for its server URL and project access. Pin the extension, connect it to that server, open a website, start review, and send only after inspecting the draft. A screenshot review collects point-in-time page diagnostics. You can turn off their inclusion before sending; start diagnostics separately if you need earlier console and network events. Recording starts through an explicit action. Optional all-website access enables right-click review; you can review from the toolbar without granting it. Browser-protected pages cannot be captured. The server operator controls access, storage and retention. Setup: https://feedbacks.softinator.ai/docs/guide/getting-started Extension guide: https://feedbacks.softinator.ai/docs/guide/chrome-extension Privacy: https://feedbacks.softinator.ai/privacy.html Source: https://github.com/Softinator-TechLabs/feedbacks-oss