mcpField
C
competitor-site-analysis

competitor-site-analysis

github.com/superampedMarketing & growth7385 installsUpdated 10w ago
Repository

About competitor-site-analysis

competitor-site-analysis is a skill for collecting publicly available information from a competitor’s website and organizing it into a five-section profile. It is suited to quick website research or as a structured first step before deeper analysis.

Key features

  1. 01Collect information from up to six website pages
  2. 02Check page availability and skip missing pages
  3. 03Organize findings into five profile sections
  4. 04Cover company, product, pricing, trust, and hiring
  5. 05Flag inaccessible or unpublished information

Use cases

  1. 01Build a quick profile of a competitor’s website
  2. 02Collect public pricing and product details
  3. 03Prepare structured inputs for further analysis

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/superamped/ai-marketing-skills --skill competitor-site-analysis

After installing, type/competitor-site-analysisin Claude Code to confirm it loaded.

Details

From the project README

Competitor Site Analysis

Usage

Use for a quick website data pull on a competitor, or as a first step before deeper analysis. Produces a structured 5-section profile from publicly available website information.

Process

Step 1: Gather Inputs

Ask the user for:

  1. Competitor name — the company to analyze
  2. Competitor URL — their homepage URL
  3. Specific sections to focus on (optional) — default: all 5
Step 2: Validate & Fetch

Confirm you have competitor name and URL. Fetch the homepage to verify it's reachable.

If the site is unreachable, behind a login wall, or returns errors — tell the user and stop.

Step 3: Fetch Website Pages

Fetch up to 6 pages. Adapt URLs based on what the site actually has — not every site uses the same paths.

PageTypical URLs to TryWhat to Extract
Homepage/Tagline, headline, value prop, target audience signals, trust badges, social proof
Pricing/pricing, /plansModel, tiers, price points, value metric, free tier, annual discount, trial, feature-by-tier gating
Features/features, /productKey features, integrations, platform capabilities
About/about, /about-us, /companyFounded, team size signals, stage/funding signals, mission
Customers/customers, /case-studiesCustomer logos, testimonials, case study count, industries served
Careers/careers, /jobs, /about/careersOpen roles by department, total headcount signals, strategic hiring patterns

If a page doesn't exist (404, redirect to homepage), skip it and note "not found" for that section. Don't guess.

Step 4: Structure Findings

Organize extracted data into 5 sections:

Section 1: Company Overview
  • Tagline (their words, not yours)
  • Founded / age signals
  • Stage / funding signals (bootstrapped, seed, Series X — based on about page, team size, office count)
  • Team size signals (exact if stated, otherwise estimate from about page)
  • Headquarters / markets
  • Primary GTM motion: PLG / sales-led / content-led / community-led (based on homepage CTA — "Start free" vs. "Book a demo" vs. both)
  • Partnerships / integrations as distribution (from features page or homepage)
Section 2: Product & Positioning
  • Homepage headline (exact quote)
  • Value proposition (what they promise in 1-2 sentences)
  • Target audience (who the site speaks to)
  • Key features (top 5-8 from features page)
  • Integrations (notable ones)
  • Product-led vs. sales-led signals (self-serve signup? demo request? both?)
  • Free trial / freemium availability
Section 3: Pricing & Feature Monetization
  • Pricing model (per-seat, usage-based, flat, freemium, custom)
  • Tier breakdown with price points
  • Value metric (what they charge per unit of)
  • Annual discount percentage
  • Free tier — what's included, what's limited
  • Enterprise tier — custom pricing or listed?

Feature monetization — if the pricing page shows feature-by-tier breakdowns:

  • Which features are free vs. gated?
  • What's the key unlock at each paid tier? (the feature that justifies the upgrade)
  • What do they gate behind enterprise? (reveals what they consider highest-value)
  • Any usage limits that force upgrades? (seats, API calls, storage, etc.)

If pricing is not public, note "not public — sales-led motion" and any signals about pricing level.

Section 4: Social Proof & Trust
  • Customer logos (notable names)
  • Testimonial highlights (1-2 strongest quotes)
  • Case study count and themes
  • Trust signals (G2 badges, security certs, compliance badges)
  • Community size (if visible — Slack, Discord, forum)
  • Social media follower counts (if visible on site)
Section 5: Hiring Signals
  • Total open roles (if listed)
  • Roles by department — engineering, sales, marketing, product, customer success, etc.
  • Notable hires or role types that signal strategic direction (e.g., "ML Engineer" = investing in AI, "Enterprise AE" = moving upmarket, "Developer Advocate" = building community)
  • Hiring velocity signals — "urgently hiring", multiple roles in one department

If no careers page exists, note "no public careers page" and move on.

Output Format

Present the 5-section structured website profile:

# Competitor Site Analysis: [Name]

**URL:** [url]
**Date:** [current date]

## 1. Company Overview
[structured findings]

## 2. Product & Positioning
[structured findings]

## 3. Pricing & Feature Monetization
[structured findings]

## 4. Social Proof & Trust
[structured findings]

## 5. Hiring Signals
[structured findings]

Rules

  • Never invent data — if a page is inaccessible or info isn't visible, mark "not found."
  • Never fetch more than 6 pages per competitor.
  • Never guess at pricing — if it's not public, say so.
  • Don't add strategic interpretation — this skill extracts, it doesn't analyze. Save strategic assessment for competitor-landscape.
  • Keep extraction factual. Quote headlines, list features, note prices. Don't editorialize.
  • If key pages (pricing, features) are missing or inaccessible, flag that the profile will have gaps.
  • If the site is heavily JS-rendered and content is sparse, suggest using Playwright MCP if available.

Comments

Markdown supported