llms.txt
# Patchrooms
> Patchrooms is a visual feedback layer for AI-built apps. It turns element-anchored reviewer comments, screenshots, voice notes, console errors, URL, viewport, browser info, and page metadata into agent-ready patch context for coding agents. Not a ticket. Patch context.
Patchrooms helps reviewers point at elements in a live app preview and leave text, screenshot, or voice feedback — no reviewer account required. Patchrooms captures the technical context automatically and exports it as agent-ready Markdown, or exposes it through an MCP server, so coding agents such as Claude Code, Cursor, Lovable, v0, Bolt, Replit Agent, Codex, Windsurf, OpenCode, and Cline can act on the feedback directly. A local mode runs with no account, no key, and no server for reviewing generated artifacts on your own machine. Free during early access.
## Core pages
- [Homepage](https://patchrooms.com/): what Patchrooms is and the 30-second demo loop
- [Launch post](https://patchrooms.com/blog/agent-ready-feedback-launch/): why feedback on AI-built apps breaks down and how patch context fixes it
- [Use cases](https://patchrooms.com/use-cases): AI artifact review, visual bug reporting, UI review, QA feedback, customer feedback, agent-ready feedback
- [Workflows](https://patchrooms.com/workflows): ways to work — review, QA test pass, auto-fix loop, propose loop, spec-writing, accumulate-and-ask
- [Integrations](https://patchrooms.com/for): per-tool workflows for Lovable, Bolt, v0, Replit Agent, Cursor, Claude Code, Codex, Windsurf, OpenCode, Cline
- [Comparisons](https://patchrooms.com/alternatives): Patchrooms vs GitHub Issues, Linear, Notion, Loom, Usersnap, BugHerd, Marker.io, Userback, Markup.io, Pastel, Feedboon, Feedbucket, Plannotator, Figma comments
- [FAQ](https://patchrooms.com/faq): the seven objections people actually have before trying Patchrooms — price, timing, trust, need — answered honestly, not argued
- [Pricing](https://patchrooms.com/pricing): free during early access
- [Security](https://patchrooms.com/security): what we store, where it lives, encryption, API keys, sub-processors, no SOC 2/ISO 27001 (and what we do instead), vulnerability reporting, prompt-injection risk via MCP report content — also machine-readable at [security.json](https://patchrooms.com/security.json)
- [Blog](https://patchrooms.com/blog): guides on agent-ready feedback and staging review workflows
## For agents
If you are an AI coding agent setting Patchrooms up for a user, start here — these pages are written for you to execute, not for a human to read.
- [For agents — overview](https://docs.patchrooms.com/for-agents/): the tree — set up the project, then run the feedback→fix loop
- [Agent self-setup](https://docs.patchrooms.com/for-agents/setup/): setup token → configure widget/channels/matching over REST → mint a narrow key → leave a feedback skill → revoke the token
- [Workflow modes](https://docs.patchrooms.com/for-agents/workflows/): review / test pass / timed auto-fix loop / propose / spec-writing / accumulate-and-ask
## Documentation
- [Docs home](https://docs.patchrooms.com/): developer documentation
- [Quickstart](https://docs.patchrooms.com/start/quickstart/): project → loader script → first report → export
- [Install the widget](https://docs.patchrooms.com/start/install/): loader vs programmatic init, all options, local mode
- [Widget API reference](https://docs.patchrooms.com/reference/widget/): Patchrooms.init options and methods
- [MCP server](https://docs.patchrooms.com/reference/mcp/): list_reports, get_report, set_status for coding agents
- [REST API](https://docs.patchrooms.com/reference/rest-api/): server-to-server access with pr_sk_ keys
- [Troubleshooting](https://docs.patchrooms.com/troubleshooting/): launcher not appearing, origin gating, key issues
## Localized pages
English documentation remains the canonical technical source. Localized discovery pages exist in Spanish, German, Brazilian Portuguese, French, Dutch, Russian, Japanese, Indonesian, Korean, Turkish, and Italian:
- [Inicio (Español)](https://patchrooms.com/es/): qué es Patchrooms — feedback visual para apps creadas con IA
- [Feedback para agentes de IA](https://patchrooms.com/es/ai-agent-feedback/): feedback humano → contexto de patch listo para agentes
- [Widget de feedback para Lovable y v0](https://patchrooms.com/es/lovable-v0-feedback-widget/): revisión de previews de Lovable, v0, Bolt, Replit
- [Feedback para Claude Code y Cursor](https://patchrooms.com/es/claude-code-cursor-feedback/): los agentes leen reportes vía Markdown/MCP
- [Start (Deutsch)](https://patchrooms.com/de/): was Patchrooms ist — visuelles Feedback für KI-gebaute Apps
- [Feedback für KI-Agenten](https://patchrooms.com/de/ai-agent-feedback/): menschliches Feedback → strukturierter Patch-Kontext für Coding Agents
- [Lovable & v0 Feedback Widget](https://patchrooms.com/de/lovable-v0-feedback-widget/): Review von Lovable-, v0-, Bolt- und Replit-Previews
- [Feedback für Claude Code & Cursor](https://patchrooms.com/de/claude-code-cursor-feedback/): Agents lesen Reports über Markdown/MCP
- [Início (Português-BR)](https://patchrooms.com/pt-br/): feedback visual para apps de IA
- [Feedback para agentes de IA](https://patchrooms.com/pt-br/ai-agent-feedback/): feedback humano → contexto de patch pronto para agentes
- [Widget de feedback para Lovable e v0](https://patchrooms.com/pt-br/lovable-v0-feedback-widget/): revisão de previews de Lovable, v0, Bolt, Replit
- [Feedback para Claude Code e Cursor](https://patchrooms.com/pt-br/claude-code-cursor-feedback/): agentes leem relatórios via Markdown/MCP
- [Accueil (Français)](https://patchrooms.com/fr/): outil de feedback visuel pour les apps créées avec l'IA
- [Feedback pour agents IA](https://patchrooms.com/fr/ai-agent-feedback/): feedback humain → contexte de patch prêt pour l'agent
- [Widget de feedback Lovable & v0](https://patchrooms.com/fr/lovable-v0-feedback-widget/): revue de previews Lovable, v0, Bolt, Replit
- [Feedback pour Claude Code & Cursor](https://patchrooms.com/fr/claude-code-cursor-feedback/): les agents lisent les rapports via Markdown/MCP
- [Start (Nederlands)](https://patchrooms.com/nl/): visuele feedback tool voor AI-apps
- [Feedback voor AI-agenten](https://patchrooms.com/nl/ai-agent-feedback/): menselijke feedback → agent-ready patch context
- [Lovable & v0 feedback widget](https://patchrooms.com/nl/lovable-v0-feedback-widget/): review van Lovable-, v0-, Bolt- en Replit-previews
- [Feedback voor Claude Code & Cursor](https://patchrooms.com/nl/claude-code-cursor-feedback/): agents lezen rapporten via Markdown/MCP
- [Главная (Русский)](https://patchrooms.com/ru/): инструмент визуального фидбэка для ИИ-приложений
- [Фидбэк для ИИ-агентов](https://patchrooms.com/ru/ai-agent-feedback/): человеческий фидбэк → готовый patch-контекст для агента
- [Виджет для Lovable и v0](https://patchrooms.com/ru/lovable-v0-feedback-widget/): ревью превью Lovable, v0, Bolt, Replit
- [Фидбэк для Claude Code и Cursor](https://patchrooms.com/ru/claude-code-cursor-feedback/): агенты читают репорты через Markdown/MCP
- [ホーム (日本語)](https://patchrooms.com/ja/): AIで作ったアプリのためのビジュアルフィードバックツール
- [AIエージェント向けフィードバック](https://patchrooms.com/ja/ai-agent-feedback/): 人のフィードバック → エージェントが読めるパッチコンテキスト
- [Lovable・v0 フィードバックウィジェット](https://patchrooms.com/ja/lovable-v0-feedback-widget/): Lovable / v0 / Bolt / Replit のプレビューをレビュー
- [Claude Code・Cursor 向けフィードバック](https://patchrooms.com/ja/claude-code-cursor-feedback/): エージェントは Markdown/MCP でレポートを読む
- [Beranda (Bahasa Indonesia)](https://patchrooms.com/id/): alat feedback visual untuk aplikasi AI
- [Feedback untuk agen AI](https://patchrooms.com/id/ai-agent-feedback/): feedback manusia → konteks patch siap agen
- [Widget feedback Lovable & v0](https://patchrooms.com/id/lovable-v0-feedback-widget/): review preview Lovable, v0, Bolt, Replit
- [Feedback untuk Claude Code & Cursor](https://patchrooms.com/id/claude-code-cursor-feedback/): agen membaca laporan via Markdown/MCP
- [홈 (한국어)](https://patchrooms.com/ko/): AI로 만든 앱을 위한 시각적 피드백 도구
- [AI 에이전트를 위한 피드백](https://patchrooms.com/ko/ai-agent-feedback/): 사람의 피드백 → 에이전트가 읽는 패치 컨텍스트
- [Lovable, v0 피드백 위젯](https://patchrooms.com/ko/lovable-v0-feedback-widget/): Lovable, v0, Bolt, Replit 프리뷰 리뷰
- [Claude Code와 Cursor를 위한 피드백](https://patchrooms.com/ko/claude-code-cursor-feedback/): 에이전트는 Markdown/MCP로 리포트를 읽음
- [Ana sayfa (Türkçe)](https://patchrooms.com/tr/): yapay zeka uygulamaları için görsel geri bildirim aracı
- [AI ajanları için feedback](https://patchrooms.com/tr/ai-agent-feedback/): insan feedback'i → ajan için hazır patch bağlamı
- [Lovable & v0 feedback widget](https://patchrooms.com/tr/lovable-v0-feedback-widget/): Lovable, v0, Bolt, Replit preview incelemesi
- [Claude Code & Cursor için feedback](https://patchrooms.com/tr/claude-code-cursor-feedback/): ajanlar raporları Markdown/MCP ile okur
- [Home (Italiano)](https://patchrooms.com/it/): strumento di feedback visivo per app create con l'IA
- [Feedback per agenti IA](https://patchrooms.com/it/ai-agent-feedback/): feedback umano → contesto di patch pronto per l'agente
- [Widget di feedback Lovable & v0](https://patchrooms.com/it/lovable-v0-feedback-widget/): review delle preview di Lovable, v0, Bolt, Replit
- [Feedback per Claude Code e Cursor](https://patchrooms.com/it/claude-code-cursor-feedback/): gli agenti leggono i report via Markdown/MCP
Every locale above also has three deeper pages at the same prefix: `/<locale>/ai-artifact-review/` (reviewing AI-generated artifacts in place), `/<locale>/visual-bug-reporting/` (click-the-bug visual bug reports), and `/<locale>/bugherd-alternative/` (honest BugHerd comparison). English canonicals: [AI artifact review](https://patchrooms.com/use-cases/ai-artifact-review/), [Visual bug reporting](https://patchrooms.com/use-cases/visual-bug-reporting/), [BugHerd alternative](https://patchrooms.com/alternatives/bugherd/).
## Key concepts
- AI-built apps: applications generated or modified by coding agents and AI app builders
- Agent-ready feedback: feedback structured so a coding agent can act on it without interpretation
- Patch context: the reviewer's comment plus the element anchor, screenshot, URL, viewport, browser, console errors, and artifact metadata needed to reproduce and fix an issue
- Element-anchored feedback: comments pinned to a stable DOM anchor, not a prose description of "the button on the left"
- Rooms: feedback grouped per artifact or preview
- Local mode: zero-network widget mode for reviewing local artifacts without an account
