agentleFS
theyoungastronautstheyoungastronauts/polaris/skills/visual-feedback/SKILL.md

visual-feedback

Skill4 starsChanged 3 months agoMITView on GitHub

Browser-annotated UI-fix workflow via the Agentation MCP.

  • Reads credentials
  • Installs packages

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use browser annotations to drive frontend fixes.

    line 10
  2. Distinctive:

    NOT for backend-only work, API logic, or non-visual changes

    line 16
  3. Very specific:

    Use agentation_watch_annotations in a loop to auto-process feedback as it arrives.

    line 111
  4. Distinctive:

    Not acknowledging annotations — the human can't tell you've seen them

    line 123

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Skill: Visual Feedback with Agentation MCP

Overview

Use browser annotations to drive frontend fixes. Humans mark up the live page, Claude picks up annotations via MCP and resolves them — no copy-paste, no describing UI bugs in words.

When to Use

  • Fixing visual/layout issues on a running web app
  • Iterating on UI with a human reviewer annotating in-browser
  • Any frontend task where "point at the broken thing" beats describing it
  • NOT for backend-only work, API logic, or non-visual changes

Setup

1. Install the package

npm install agentation -D

2. Add the browser component

React / Next.js — in your root layout (e.g., src/app/layout.tsx):

import { Agentation } from "agentation";

// Inside <body>, after your content:
{process.env.NODE_ENV === "development" && <Agentation />}

Astro — create a React island wrapper:

// src/components/dev/AgentationOverlay.tsx
import { Agentation } from "agentation";
export default function AgentationOverlay() {
  return <Agentation />;
}

Then include in your base layout (e.g., src/layouts/BaseLayout.astro):

---
import AgentationOverlay from '@/components/dev/AgentationOverlay.tsx';
---
<!-- Inside <body>, after <Footer /> -->
{import.meta.env.DEV && <AgentationOverlay client:only="react" />}

Astro requires @astrojs/react, react, and react-dom as dependencies.

3. Configure MCP

Option A — project .mcp.json (recommended, shared with team):

{
  "mcpServers": {
    "agentation": {
      "command": "npx",
      "args": ["agentation-mcp", "server"]
    }
  }
}

Option B — per-user CLI:

claude mcp add agentation -- npx agentation-mcp server

4. Verify

npx agentation-mcp doctor

The browser toolbar connects to an HTTP server (port 4747 default), which feeds annotations to Claude via MCP stdio.

MCP Tools Available

ToolPurpose
agentation_list_sessionsList active annotation sessions
agentation_get_sessionGet annotations for a session
agentation_get_pendingPending annotations for a session
agentation_get_all_pendingAll pending across sessions
agentation_acknowledgeMark annotation as seen
agentation_resolveMark annotation as fixed
agentation_dismissDismiss an annotation
agentation_replyReply to an annotation thread
agentation_watch_annotationsBlock until new annotations arrive

Workflow

  1. Human opens the running app with the Agentation toolbar active
  2. Human clicks elements and leaves annotations (fix / change / question / approve)
  3. Annotations include CSS selectors and React component paths — use these to locate code
  4. Fetch pending annotations, acknowledge them, make the fix
  5. Mark resolved once the change is committed or verified

Hands-Free Mode

Use agentation_watch_annotations in a loop to auto-process feedback as it arrives. Good for rapid iteration sessions where the human is continuously reviewing.

Annotation Data

Each annotation includes:

  • Element selector — CSS selector for the annotated DOM element
  • Component path — React component tree path (when available)
  • Intent — fix, change, question, or approve
  • Severity — blocking, important, or suggestion
  • Comment — Human's description of the issue

Common Mistakes

  • Ignoring severity — address blocking annotations before suggestion
  • Not acknowledging annotations — the human can't tell you've seen them
  • Resolving without verifying — mark resolved only after the fix is confirmed
  • Using watch mode without a batch window — set a reasonable timeout to avoid thrashing

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.