feat: Add Review Analyzer kit [agentkit-challenge]

kits/review-analyzer/.env.example

@@ -0,0 +1,4 @@
+REVIEW_ANALYZER_FLOW_ID = "YOUR_FLOW_ID"
+LAMATIC_API_URL = "YOUR_LAMATIC_API_URL"
+LAMATIC_PROJECT_ID = "YOUR_LAMATIC_PROJECT_ID"
+LAMATIC_API_KEY = "YOUR_LAMATIC_API_KEY"

kits/review-analyzer/README.md

@@ -0,0 +1,94 @@
+# Review Analyzer Kit
+
+An AI-powered Chrome Extension that scrapes product reviews from e-commerce sites (like Amazon) and uses [Lamatic.ai](https://lamatic.ai) to synthesize customer consensus, extract key pros/cons, and calculate a Trust Score to identify fake or low-effort reviews.
+
+## 🚀 Overview
+Review Analyzer is built with Next.js and Tailwind CSS, communicating with a Lamatic AI flow backend. By loading the Chrome Extension, users can trigger review scraping with a single click. The scraped data is securely passed to a Next.js server, which runs the reasoning agent workflow on Lamatic AI to generate the summary dashboard.
+
+### 🎯 The Problem (AgentKit Challenge)
+Online shoppers are overwhelmed by hundreds of reviews per product, many of which are fake, low-effort, or artificially inflated. It takes too much time to read through them to find the true consensus, pros, and cons. 
+
+### 💡 The Value & Solution
+This extension solves the problem by providing a simple, one-click solution. It scrapes the reviews and uses a Lamatic AI agent to:
+- Save time by summarizing the overall sentiment.
+- Extract a clean list of the top Pros and Cons.
+- Calculate a **Trust Score (0-100)** to identify suspicious, fake, or duplicate reviews.
+This makes e-commerce research faster, safer, and entirely automated.
+
+## 🛠️ Folder Structure
+```
+kits/review-analyzer/
+├── lamatic.config.ts         # Metadata definition
+├── config.json               # Legacy config to pass CI validation
+├── agent.md                  # Agent capability & setup document
+├── README.md                 # This guide
+├── constitutions/
+│   └── default.md            # LLM behavior guide
+├── flows/
+│   └── review-analyzer.ts    # Flow config exported from Lamatic Studio
+└── apps/                     # Next.js web application & Chrome Extension
+    ├── package.json          # Next.js dependencies
+    ├── app/                  # Next.js page router (iframe popup view)
+    ├── lib/                  # Library files (lamatic client)
+    ├── actions/              # Server Actions calling Lamatic flow
+    └── extension/            # Chrome Extension (manifest, scripts, popup iframe)
+```
+
+---
+
+## 🔑 Setup & Installation
+
+### Step 1: Clone the Repo & Install Dependencies
+First, clone your fork and install Next.js dependencies:
+```bash
+cd kits/review-analyzer/apps
+npm install
+```
+
+### Step 2: Set up Environment Variables
+Create a `.env.local` file inside the `apps` directory with the following keys:
+```bash
+# Lamatic Flow ID
+REVIEW_ANALYZER_FLOW_ID="your-flow-id"
+
+# Lamatic Credentials (Get from Settings in Lamatic Studio)
+LAMATIC_API_URL="https://studio.lamatic.ai/api/your-project-url"
+LAMATIC_PROJECT_ID="your-project-id"
+LAMATIC_API_KEY="your-api-key"
+```
+
+### Step 3: Run the Next.js App
+Start the Next.js server locally to act as the backend API bridge for the Chrome Extension:
+```bash
+npm run dev
+```
+The Next.js dashboard will be running at `http://localhost:3000`.
+
+### Step 4: Install the Chrome Extension
+1. Open Google Chrome and navigate to `chrome://extensions/`.
+2. Enable **Developer mode** using the toggle switch in the top-right corner.
+3. Click the **Load unpacked** button in the top-left corner.
+4. Select the `kits/review-analyzer/apps/extension` folder from your local directory.
+5. The extension is now active in your browser.
+
+---
+
+## 🧠 Lamatic Flow Workflow Setup
+
+In the [Lamatic Cloud Studio](https://studio.lamatic.ai):
+1. Create a new flow named `review-analyzer`.
+2. **Define the Input**: Set the trigger to API (GraphQL). The input schema should take an array of strings (e.g. `reviews: [String]`).
+3. **Add LLM Node**: Use Claude 3.5 Sonnet or GPT-4o. Set the system prompt to analyze the reviews, compile a consensus summary, isolate key pros & cons, and detect suspicious patterns (such as exact text duplication, excessively short phrasing, or repetitive reviews) to assign a **Trust Score (0 to 100)**.
+4. **Define the Output**: Map the output to a GraphQL schema returning the following fields:
+   - `summary`: string
+   - `pros`: string[]
+   - `cons`: string[]
+   - `trustScore`: int
+   - `trustLabel`: string
+   - `analysisDetail`: string
+5. **Deploy**: Deploy the flow to retrieve your Flow ID and Project details, and populate `.env.local`.
+
+---
+
+## 🤝 Contributing
+Feel free to open issues or PRs! To submit your changes to the AgentKit repo, ensure your PR is labeled with `agentkit-challenge`.

kits/review-analyzer/agent.md

@@ -0,0 +1,58 @@
+# Agent Identity: Review Analyzer
+
+## Overview
+Review Analyzer is an AI-powered assistant designed to help e-commerce shoppers make informed decisions instantly. By scraping and analyzing hundreds of reviews from product pages, it delivers a clear consensus, highlights critical pros & cons, and calculates a Trust Score to identify fake or sponsored comments.
+
+## Purpose
+The sheer volume of reviews on sites like Amazon makes it impossible for shoppers to read everything, and many reviews are generated by bots, incentivized programs, or contains low-effort filler. Review Analyzer:
+- **Saves Time**: Aggregates and synthesizes hundreds of reviews into a few readable sentences.
+- **Exposes Hidden Flaws**: Isolates negative feedback that might be buried under five-star ratings.
+- **Protects from Fraud**: Audits reviews for bot-like phrasing, repetitive patterns, or sponsored templates to score review trust.
+
+## Flows
+
+### 1. `review-analyzer` Flow
+This is the primary flow triggered by the Chrome Extension. It takes scraped reviews, processes them using a large language model (Claude 3.5 Sonnet or GPT-4o) through Lamatic AI, and returns a structured analysis.
+
+- **Trigger**: API Trigger (GraphQL execution).
+- **Processing**:
+  - Summarizes the overall customer sentiment and consensus.
+  - Extracts lists of distinct Pros and Cons.
+  - Examines individual review properties (length, repetitive words, generic bot-like phrasing) to calculate an aggregate Trust Score.
+- **Output**: JSON payload returning:
+  - `summary`: The general consensus summary.
+  - `pros`: A list of product benefits.
+  - `cons`: A list of product drawbacks or complaints.
+  - `trustScore`: Score from 0 to 100 indicating review reliability.
+  - `trustLabel`: Sentiment classification of trust (e.g. "Highly Trusted", "Needs Review", "Unreliable").
+  - `analysisDetail`: Explanation of why the trust score was assigned.
+
+## Guardrails
+- **Input Limit**: The flow handles a maximum of 50 reviews per call to optimize response time and LLM token limits.
+- **Content Security**: Content is filtered to strip HTML tags and scripts before LLM processing to prevent prompt injection.
+- **Language Boundaries**: If reviews are in languages other than English, the LLM will automatically translate the summary and pros/cons into English while processing.
+
+## Environment Setup
+The Next.js bridge app requires the following environment variables:
+
+| Variable Name | Description | Source |
+|---|---|---|
+| `REVIEW_ANALYZER_FLOW_ID` | Flow ID of the deployed Lamatic flow | Lamatic Studio Flow page |
+| `LAMATIC_API_URL` | Endpoint URL for the Lamatic project | Lamatic Studio API Settings |
+| `LAMATIC_PROJECT_ID` | Project ID in Lamatic | Lamatic Studio Settings |
+| `LAMATIC_API_KEY` | Secret API key for authentication | Lamatic Studio API Settings |
+
+## Quickstart
+1. Set up the environment variables in `apps/.env.local`.
+2. Deploy your workflow in the Lamatic AI Studio and export it to `flows/review-analyzer.ts`.
+3. Run `npm install` and `npm run dev` in the `apps/` directory.
+4. Load the Chrome Extension folder `apps/extension` into your Chrome browser via `chrome://extensions`.
+5. Visit any Amazon or e-commerce product page, click the extension icon, and view the analysis.
+
+## Common Failure Modes
+
+| Symptom | Cause | Resolution |
+|---|---|---|
+| Trust Score is always 100 | LLM prompt failed to detect patterns or reviews are too sparse | Ensure the scraped reviews are correctly parsed and passed. Add more diverse reviews to the page. |
+| Extension popup shows connection error | Next.js server is not running or env variables are incorrect | Ensure Next.js is running at the configured URL (`http://localhost:3000`) and the API credentials are valid. |
+| Scraper yields empty results | Target site updated their DOM/classnames | Update the selectors in `content.js` to match the new review container class names on the target e-commerce site. |

kits/review-analyzer/apps/.env.example

@@ -0,0 +1,4 @@
+REVIEW_ANALYZER_FLOW_ID = "YOUR_FLOW_ID"
+LAMATIC_API_URL = "YOUR_LAMATIC_API_URL"
+LAMATIC_PROJECT_ID = "YOUR_LAMATIC_PROJECT_ID"
+LAMATIC_API_KEY = "YOUR_LAMATIC_API_KEY"

kits/review-analyzer/apps/.gitignore

@@ -0,0 +1,29 @@
+# See https://help.github.com/articles/ignoring-files/ for more about ignoring files.
+
+# dependencies
+/node_modules
+
+# next.js
+/.next/
+/out/
+
+# production
+/build
+
+# debug
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+.pnpm-debug.log*
+
+# env files
+.env
+.env.local
+.env.*.local
+
+# vercel
+.vercel
+
+# typescript
+*.tsbuildinfo
+next-env.d.ts

kits/review-analyzer/apps/.npmrc

@@ -0,0 +1 @@
+legacy-peer-deps=true

kits/review-analyzer/apps/README.md

@@ -0,0 +1,155 @@
+# Reddit Scout by Lamatic.ai
+
+<p align="center">
+  <a href="https://reddit-scout-tawny.vercel.app/" target="_blank">
+    <img src="https://img.shields.io/badge/Live%20Demo-black?style=for-the-badge" alt="Live Demo" />
+  </a>
+</p>
+
+**Reddit Scout** is an AI-powered product review research tool built with [Lamatic.ai](https://lamatic.ai). It searches Reddit for real user opinions and generates structured, scannable review summaries for any product or topic.
+
+[![Deploy with Vercel](https://vercel.com/button)](https://vercel.com/new/clone?repository-url=https://github.com/Lamatic/AgentKit&root-directory=kits/agentic/reddit-scout&env=REDDIT_SCOUT_FLOW_ID,LAMATIC_API_URL,LAMATIC_PROJECT_ID,LAMATIC_API_KEY&envDescription=Your%20Lamatic%20Reddit%20Scout%20keys%20are%20required.)
+
+---
+
+## The Problem
+
+When researching a product before buying, the most honest opinions live on Reddit — not in influencer reviews or star ratings. But finding and reading through dozens of Reddit threads is time-consuming. People manually Google "site:reddit.com [product] reviews", open 10 tabs, read hundreds of comments, and try to mentally synthesize what the consensus is. This takes 15-20 minutes per product and is hard to do consistently.
+
+**Nobody has built a focused tool that automates this specific workflow.**
+
+## The Approach
+
+Reddit Scout is a single-purpose tool: type a product name, get a structured review summary from Reddit.
+
+**How it works:**
+
+1. User enters a product or topic (e.g., "HP Victus", "Sony WH-1000XM5")
+2. An LLM generates Reddit-scoped search queries
+3. Google Serper API finds the most relevant Reddit threads
+4. Serper's scrape endpoint extracts full thread content (posts + comments)
+5. A second LLM analyzes all content and produces a structured summary
+
+**Tech stack:**
+- **Lamatic AI** — orchestrates the entire flow (search, scrape, analyze)
+- **Google Serper API** — Reddit-scoped web search and thread scraping
+- **Next.js** — clean, fast frontend
+- **ReactMarkdown** — renders AI output as formatted markdown
+
+## The Result
+
+A clean, single-page app where users type a product name and get a structured review in seconds:
+
+- **Overall Sentiment** — positive/negative split at a glance
+- **What Users Love** — key pros from real users
+- **Common Complaints** — honest downsides
+- **Notable User Quotes** — direct quotes with context
+- **Frequently Discussed Features** — what keeps coming up
+- **Final Verdict** — direct recommendation
+
+**Live demo:** [reddit-scout-tawny.vercel.app](https://reddit-scout-tawny.vercel.app/)
+
+## Tradeoffs & Assumptions
+
+- **Scrape vs. full crawl:** We use Serper's scrape endpoint (single-page) instead of Firecrawl (which doesn't support Reddit). This means we get thread content but not all nested comments — we capture the most relevant discussion.
+- **10 threads per search:** We scrape the top 10 Reddit results. This balances speed (~20s) with coverage. More threads would mean longer wait times.
+- **No caching:** Every search hits the API fresh. Adding a cache layer would reduce costs for repeated queries but wasn't needed for this scope.
+- **Single product search:** No comparison mode (e.g., "X vs Y") — keeping the scope narrow and the output clean.
+
+---
+
+## Lamatic Setup
+
+Before running this project, you must build and deploy the flow in Lamatic, then wire its config into this codebase.
+
+**Pre: Build in Lamatic**
+1. Sign in or sign up at https://lamatic.ai
+2. Create a project
+3. Click "+ New Flow" and select "Flow"
+4. Build the Reddit Scout flow:
+   - Trigger Node (GraphQL) with input `{ "query": "string" }`
+   - LLM Node: Generate Reddit search query
+   - Web Search Node (Serper): Search Google scoped to Reddit
+   - Code Node: Extract URLs from search results
+   - Batch Node: Iterate over URLs
+   - Web Search Node (Serper Scrape): Scrape each Reddit thread
+   - Code Node: Extract all scraped text
+   - LLM Node: Analyze and generate structured review summary
+   - Response Node: Return with output mapping `{ "answer": "{{LLMNode.output.generatedResponse}}" }`
+5. Configure providers and API credentials (Serper API key)
+6. Deploy the flow in Lamatic and obtain your .env keys
+
+**Post: Wire into this repo**
+1. Create a `.env` file and set the keys
+2. Install and run locally:
+   - `npm install`
+   - `npm run dev`
+3. Deploy (Vercel recommended):
+   - Import your repo, set the project's Root Directory to `kits/agentic/reddit-scout`
+   - Add env vars in Vercel (same as your `.env`)
+   - Deploy and test your live URL
+
+---
+
+## Key Features
+- Searches Reddit for product reviews using Google Serper API
+- Scrapes full Reddit thread content via Serper scrape endpoint
+- AI-generated structured review summaries with sentiment analysis
+- Pros/cons extraction and notable user quotes
+- Clean markdown rendering of results
+
+---
+
+## Required Keys and Config
+
+| Item | Purpose | Where to Get It |
+|------|---------|-----------------|
+| Lamatic API Key | Authentication for Lamatic AI APIs | [lamatic.ai](https://lamatic.ai) |
+| Serper API Key | Google Search API for Reddit search | [serper.dev](https://serper.dev) |
+
+### 1. Environment Variables
+
+Create `.env.local` with:
+
+```bash
+REDDIT_SCOUT_FLOW_ID = "YOUR_FLOW_ID"
+LAMATIC_API_URL = "YOUR_API_ENDPOINT"
+LAMATIC_PROJECT_ID = "YOUR_PROJECT_ID"
+LAMATIC_API_KEY = "YOUR_API_KEY"
+```
+
+### 2. Install & Run
+
+```bash
+npm install
+npm run dev
+# Open http://localhost:3000
+```
+
+---
+
+## Repo Structure
+
+```
+/actions
+ └── orchestrate.ts        # Lamatic workflow orchestration
+/app
+ └── page.tsx              # Main search form UI
+/components
+ ├── header.tsx            # Header component
+ └── ui                    # shadcn/ui components
+/lib
+ └── lamatic-client.ts     # Lamatic SDK client
+/public
+ └── lamatic-logo.png      # Lamatic branding
+/flows
+  └── reddit-scout/        # Exported Lamatic flow
+/package.json              # Dependencies & scripts
+```
+
+---
+
+## License
+
+MIT License
+