Back to blog

Automate competitive research with Claude Code

by Mack 4 min read claude-codecompetitive-researchskills

The problem with competitive research

Most people do it twice a year, badly. They spend an afternoon on a competitor’s site, take some notes, and call it done.

The notes go stale. The competitor ships a new pricing tier. A new player enters the market. You miss it.

The alternative is hiring someone to track it regularly. Or building a spreadsheet no one updates. Neither one works long-term.

What works is a repeatable system you can run whenever you need fresh intel.

What Competitive Radar does

Competitive Radar is a Claude Code skill. Give it a company domain or name. It maps the site, scrapes the pages that matter, and returns a structured intelligence brief.

One command. Full picture.

Here is what the output covers:

  • Company overview and what they sell
  • Pricing model and published tiers
  • Target market and customer evidence (logos, case studies, personas mentioned)
  • Team and leadership signals
  • Hiring patterns (what roles are open reveals where a company is investing)
  • Market positioning in their own words
  • Gaps: what the site is silent about, where their positioning is weak
  • A list of every page reviewed so you can verify the sourcing

That last one matters. Competitive Radar cites its sources. Every claim traces back to a specific URL. No fabricated stats, no stale training data presented as current fact.

How it works

The skill runs in three stages.

First, it maps the site. It collects the full URL structure from the target domain, up to 100 pages. This step is fast because it is not reading content yet.

Second, it filters. High-signal pages get prioritized: /pricing, /about, /products, /careers, /case-studies. Low-signal pages get skipped: privacy policies, sitemaps, paginated archives. The skill reads the 15-20 most valuable pages.

Third, it synthesizes. Claude reads the scraped content and builds the structured brief. The output is markdown, clean, ready to act on.

The whole process takes two to four minutes depending on the target site’s size.

A concrete example

Say you are building a project management tool for freelancers and you want to understand how Basecamp positions itself right now.

You run:

competitive-radar basecamp.com

The skill maps the site, pulls /pricing, /about, the product pages, recent blog posts, and the careers section. It comes back with something like this:

# Company Profile: Basecamp

**URL:** basecamp.com
**Analyzed:** 2026-04-08
**Pages Reviewed:** 17

## Products / Services
Basecamp is an all-in-one project management tool combining
to-dos, messaging, file storage, and scheduling. Single-product company.
One flat price: $99/month for unlimited users (no per-seat pricing).

## Target Market
Small businesses, agencies, remote teams. Explicitly anti-enterprise.
"Calm company" positioning recurs across multiple pages.

## Pricing Intelligence
Flat $99/month. Free tier for personal projects (3 projects, 1GB storage).
No upsell tiers. Strong messaging against per-seat pricing models.

## Weaknesses / Gaps
No native time tracking. No client billing. No integrations marketplace
featured prominently. Freelancers with billing needs are underserved.

## Hiring
3 open roles: one senior designer, two customer support.
No engineering hiring at current snapshot. Suggests stable product phase.

You now know the gap: billing and time tracking for freelancers. Basecamp does not serve that need and does not appear to be moving toward it. You have a positioning angle, grounded in evidence, in four minutes.

The --topic flag

If you want depth on one area, add a topic flag:

competitive-radar basecamp.com --topic pricing

This adds a dedicated pricing deep-dive: comparison tables, hidden costs, discount signals, free tier mechanics. Useful before a sales call or a pricing decision of your own.

Other flags: --topic team, --topic products, --topic investors, --topic hiring.

What you get versus what you do not

Competitive Radar returns a brief you can act on in an hour, not a 50-page report you will never finish reading.

It is grounded in the target’s current web presence. If a company has not published something publicly, the skill will say so rather than speculate. The “Intelligence Gaps” section tells you exactly where to look next: LinkedIn for team depth, Crunchbase for funding history, G2 for customer reviews.

It is not a substitute for talking to actual customers or getting on sales calls. It is the layer of research you should always have done but usually skipped.

How to install it

The skill is free on GitHub.

git clone https://github.com/RealityResearch/skills

Then in Claude Code:

/add-skill skills/competitive-radar/SKILL.md

Run it:

competitive-radar [domain or company name]

That is the full setup. No API keys beyond what Claude Code already uses.

Where this fits in the library

Competitive Radar is one of four skills currently published. The others are Autoclip, Content Engine, and Audience Finder. New skills ship weekly.

Each skill works standalone. But they compose. Competitive Radar tells you where the gap is. Audience Finder tells you who to reach. Content Engine builds the posts. You can run the whole research-to-content loop in under an hour.

That chain is what separates one-off prompting from a system. The skills are the first layer.

All of our tools are free on GitHub. The full system is $497 lifetime at realityresearch.studio.