Item logo image for AgentsRoom Feedback

AgentsRoom Feedback

agentsroom.dev
ExtensionDeveloper Tools10 users
Item media 4 (screenshot) for AgentsRoom Feedback
Item media 1 (screenshot) for AgentsRoom Feedback
Item media 2 (screenshot) for AgentsRoom Feedback
Item media 3 (screenshot) for AgentsRoom Feedback
Item media 4 (screenshot) for AgentsRoom Feedback
Item media 1 (screenshot) for AgentsRoom Feedback
Item media 1 (screenshot) for AgentsRoom Feedback
Item media 2 (screenshot) for AgentsRoom Feedback
Item media 3 (screenshot) for AgentsRoom Feedback
Item media 4 (screenshot) for AgentsRoom Feedback

Overview

Send bugs, features and questions from any website to an AgentsRoom backlog. Point at any element on a page to attach context.

AgentsRoom Feedback is a bug reporting and feedback extension for web developers, product teams and QA testers. In plain words: this extension lets you report a bug, a visual glitch, a typo, a feature request or any feedback directly from the webpage you are looking at. Click the toolbar icon, type what is wrong, optionally pin the exact element on the page and attach a screenshot, then hit send. The feedback is saved as a ticket in your AgentsRoom (MAC OS APP) project backlog (or in a public backlog of your choice), with the page URL, the page title, the selected text, a CSS selector for the clicked element and the page HTML context automatically attached. In one sentence: AgentsRoom Feedback turns any webpage into a structured bug report or feature request and sends it to your AgentsRoom backlog in one click, so your team (or your AI coding agents) can pick it up and fix it. No more screenshots pasted into Slack. No more "I cannot repro" pings. No more half described bugs rotting in a Notion page nobody opens. ========================================== WHAT IT DOES ========================================== AgentsRoom Feedback is the official Chrome companion extension for AgentsRoom, the visual multi agent command center for Claude Code, Codex, Gemini, Aider and OpenCode. It lets you, your teammates and your clients file actionable tickets from any webpage in seconds. Three ways to send feedback, one goal: turn raw observations into work your team (or your AI agents) can actually execute. 1. SIGNED IN ACCOUNT MODE Log in with your AgentsRoom account and pick one of your projects from the dropdown. Every ticket you file is created directly in your own backlog, attached to the correct project, with the page URL, title and selected text automatically captured. You can also attach a screenshot (PNG, JPG or WEBP, up to 4 MB) that is persisted alongside the ticket so you never lose visual context. 2. PUBLIC BACKLOG MODE Paste a public backlog slug (for example "my-project") and start submitting anonymous feedback to that backlog. Perfect for shipping a beta to testers, collecting feedback from clients who do not have an AgentsRoom account, or letting an entire community report issues on a landing page without any friction. 3. POINT ON PAGE MODE The killer feature. Click "Point on page", hover any element on the current website, click to pin it, and a comment bubble pops up anchored exactly where you clicked. Write what is wrong, hit send, and the extension automatically: - generates a ticket title from your comment (powered by an LLM via the AgentsRoom API) - captures the clicked element's outer HTML (up to 1.5 KB) - generates a readable CSS selector so your team can locate the exact DOM node - attaches the page URL and viewport info - routes the ticket to the default agent configured for that backlog The result: instead of "button is broken on the pricing page", you receive "Pricing CTA button misaligned on mobile breakpoint, selector: .pricing-section > .cta-primary, HTML snippet included". That is the difference between a ticket that rots and a ticket that ships. ========================================== WHY YOU WANT THIS ========================================== If you are using AgentsRoom to run AI coding agents (Claude Code, Codex, Gemini, Aider, OpenCode) on your codebase, the bottleneck is rarely the agent. The bottleneck is the loop between seeing a problem and describing it well enough for the agent to fix it without asking five follow up questions. AgentsRoom Feedback collapses that loop from minutes to seconds. - Designers file visual bugs with pinpoint accuracy instead of annotated Figma exports. - Product managers capture "this copy is wrong" while browsing the staging environment. - QA engineers reproduce issues and ship them straight into the backlog with a selector, a screenshot and a URL. - Clients submit feedback without learning your tooling, without creating an account, without writing Markdown. - Developers triage faster because every ticket arrives with the context needed to take action. ========================================== HOW TO USE IT WITH AGENTSROOM ========================================== Step 1. Install AgentsRoom Download the AgentsRoom desktop app from https://agentsroom.dev. It is a free native macOS app that lets you run up to 14 AI coding agents in parallel on your repository, with full support for Claude Code, Codex, Gemini, Aider and OpenCode, a visual board, shared context and real time terminal streaming. Pick the provider you prefer per agent, or mix and match across a single project. Step 2. Create a project and a backlog Open AgentsRoom, create a project, point it at your git repository. Every project comes with a built in backlog: a Kanban like view where you drop tickets, assign them to agents (frontend, backend, QA, docs, copywriter, and more), and watch Claude Code, Codex, Gemini, Aider or OpenCode pick them up and work through them. Step 3. Install the Chrome extension Install AgentsRoom Feedback from the Chrome Web Store. Click the toolbar icon, sign in with the same credentials you use for the desktop app, and pick your project from the dropdown. That is all the setup you need. Step 4. Browse and file tickets Open the website you are building or testing. Whenever you spot something worth fixing: - open the popup and type a quick note, or - click "Point on page" and pin the exact element that is broken. Your ticket instantly appears in the backlog inside the AgentsRoom desktop app, with the page URL, screenshot (optional), selected text and DOM context already filled in. Step 5. Let your team (or your agents) do the work From the desktop app, drag the new ticket onto an agent, or let the default agent pick it up automatically. Your chosen provider (Claude Code, Codex, Gemini, Aider or OpenCode) reads the context you captured, locates the relevant files in your repo, proposes a patch, runs your tests, and pings you when the fix is ready for review. Step 6. Share the backlog with clients Every backlog can be made public or unlisted via the "Share backlog" modal in the desktop app. Send the public URL to your client, they install the Chrome extension, paste the slug, and start sending you feedback directly. You can even configure a default agent for incoming public tickets so that low effort reports get a first pass from Claude Code, Codex, Gemini, Aider or OpenCode before you even look at them. ========================================== FEATURES IN DETAIL ========================================== AUTO CAPTURE OF PAGE CONTEXT Every ticket, regardless of the mode, automatically captures the current page URL, the page title, any text you had selected at the moment you opened the popup, and the active browser tab's window size. You never have to copy paste any of that yourself. QUOTED SELECTION PRE FILL Select a paragraph before opening the popup and it is auto quoted in the ticket description. Great for typos, wrong copy or translation issues, no manual retyping. PIN AND COMMENT (POINT ON PAGE) Pointer mode is implemented as a content script that injects a Shadow DOM widget on top of the page, so it never interferes with the site's own styles or JavaScript. Hover highlights elements, a click pins an anchor exactly where you clicked, and a floating bubble lets you type your comment inline. The UI is keyboard friendly (Escape to cancel, Enter to submit) and accessible. SMART TITLE GENERATION The extension calls POST /api/backlog/generate-title to ask an LLM for a concise, actionable ticket title based on your comment and the captured context. If the API is slow or unreachable, a local fallback generates a decent title from the first words of your comment so you never get stuck. SCREENSHOT ATTACHMENTS Attach a PNG, JPG or WEBP screenshot up to 4 MB. Screenshots are persisted as a BacklogImage record and displayed inline in the desktop app's ticket detail view. Agents can read the image metadata and, with vision enabled models, interpret the screenshot directly. DOM CONTEXT FOR TRIAGE When you use point on page mode, the extension captures the target element's outer HTML (truncated to 1.5 KB) and a readable CSS selector. That snippet is embedded in the ticket body so your frontend team can immediately grep the codebase for the right component, without having to ask "where in the tree is this?". DEFAULT AGENT ROUTING Each backlog can define a default agent that picks up new tickets (configured in the "Share backlog" modal of the desktop app). Point on page tickets automatically inherit that routing, so feedback from clients lands in the hands (or rather, the context window) of the right specialist without any manual triage. JOIN INVITED PROJECTS Project owners can share a backlog slug with collaborators who do not already have access. In the extension, click "Join a project you were invited to", paste the slug, and the extension probes GET /api/chrome-extension/check-slug to figure out whether the backlog is public, unlisted or restricted. Public and unlisted backlogs are added to your dropdown instantly. Restricted backlogs tell you to ask the owner for access. REMEMBER ME, BUT SAFELY The extension remembers your email after sign in and optionally your password (if you opt in) in chrome.storage.local, sandboxed per extension ID. On 401, it silently re authenticates. Uncheck the box on shared computers. Sign out clears the token and password but keeps the email for one click re sign in. KEYBOARD SHORTCUTS Tab through the form, Cmd Enter to submit, Escape to cancel point on page mode. The popup is designed to be used without ever touching the mouse once you have picked a project. ========================================== USE CASES ========================================== SOLO BUILDER WITH AI CODING AGENTS You are building a SaaS on your own. You browse your own staging site, spot issues, file them with point on page, and let your AgentsRoom agents (Claude Code, Codex, Gemini, Aider or OpenCode, your pick) implement the fix while you move on to the next page. Zero context switching. AGENCY WORKING FOR CLIENTS You ship a staging URL to a client. Instead of getting vague emails like "the form is weird", you get a ticket with a pinned element, a screenshot, a URL and a CSS selector. Your team fixes it before the next standup. QA TEAM Testers run through a test plan on a staging environment. Every bug becomes a ticket in under ten seconds, with selector, screenshot and reproduction URL included by default. PRODUCT MANAGER You walk through the product every morning, pin UX issues with point on page, and let your frontend team iterate on the fixes overnight. DESIGNER REVIEWING AN IMPLEMENTATION You compare the live site to the Figma file, point on each pixel that is off, attach the Figma frame as a screenshot, and assign the ticket to the frontend team with precise redlines. ========================================== PRIVACY AND PERMISSIONS ========================================== AgentsRoom Feedback requests only the minimum permissions needed to do its job: - activeTab: to read the current tab's URL and the text the user has selected at the moment the popup is opened. The extension never reads pages in the background. - scripting: to inject the tiny selection capture script and, in point on page mode, the content picker overlay. Scripts only run on the tab you explicitly interact with. - storage: to remember your session token, your last used public slug, your joined public backlogs and optionally your password (only if you opt in via the checkbox). All data is stored in chrome.storage.local, sandboxed per extension ID. - host permissions for agentsroom.dev, *.backlog.agentsroom.dev and http://localhost:2203 (dev builds only): to POST tickets and authenticate against the AgentsRoom API. The extension does NOT: - track your browsing history - send analytics or telemetry - read the content of pages you do not explicitly interact with - share any data with third parties - include ads The only network calls made by the extension are: - POST /api/chrome-extension/auth (sign in) - GET /api/chrome-extension/projects (list your projects) - GET /api/chrome-extension/check-slug (probe a public slug) - POST /api/chrome-extension/tickets (create a ticket in a signed in account) - POST /api/embed/{slug}/submit (create a ticket in a public backlog) - POST /api/backlog/generate-title (ask an LLM for a ticket title) All of these calls target agentsroom.dev (or localhost in dev builds). Your feedback goes to your backlog, period. ========================================== ABOUT AGENTSROOM ========================================== AgentsRoom is a visual multi agent command center for AI coding CLIs. It supports Claude Code, Codex, Gemini, Aider and OpenCode out of the box, and lets developers, indie hackers, agencies and teams run multiple AI coding agents in parallel on their own codebase, with a shared context, a visual task board, real time terminal streaming, and a preview tunnel for non technical reviewers. Key AgentsRoom features: - Multi provider support: Claude Code, Codex, Gemini, Aider and OpenCode, switchable per agent at any time. - 14 specialized agent roles (frontend, backend, QA, DevOps, copywriter, designer, SEO, security, and more), each with its own system prompt tuned for its specialty. - Visual Kanban backlog with drag and drop task assignment. - Native macOS desktop app built on Electron, with xterm.js powered terminals for each agent. - Companion iOS and Android mobile apps to monitor your agents on the go. - End to end encrypted relay (E2EE) so that desktop, mobile and web clients stay in sync without any server ever seeing your code or conversations. - Prompt library, CLAUDE.md editor, preview tunnel, share backlog, remote triggers, cron schedules and much more. The desktop app is a free download at https://agentsroom.dev. The Chrome extension is the last mile of the feedback loop: it takes what you see in your browser and turns it into a structured ticket in your backlog. ========================================== SUPPORT AND LINKS ========================================== Website: https://agentsroom.dev Desktop app download: https://agentsroom.dev Contact: support@agentsroom.dev If you run into any issue with the extension, file a ticket using the extension itself against the AgentsRoom public backlog and we will get back to you quickly. Install AgentsRoom Feedback, point on your staging site, and ship faster. That is the whole pitch.

Details

  • Version
    0.2.1
  • Updated
    April 27, 2026
  • Size
    38.65KiB
  • Languages
    English
  • Developer
    Website
    Email
    chauvin.jbaptiste@gmail.com
  • Non-trader
    This developer has not identified itself as a trader. For consumers in the European Union, please note that consumer rights do not apply to contracts between you and this developer.

Privacy

Manage extensions and learn how they're being used in your organization

AgentsRoom Feedback has disclosed the following information regarding the collection and usage of your data. More detailed information can be found in the developer's privacy policy.

AgentsRoom Feedback handles the following:

Personally identifiable information
User activity

This developer declares that your data is

  • Not being sold to third parties, outside of the approved use cases
  • Not being used or transferred for purposes that are unrelated to the item's core functionality
  • Not being used or transferred to determine creditworthiness or for lending purposes

Support

For help with questions, suggestions, or problems, visit the developer's support site

Google apps