competitor-profiling turns competitor research into a repeatable, dated profile. It selects a data stack, saves raw scrapes, SEO data and reviews to disk, then writes a profile from a fixed template and a cross-competitor summary.
Every claim is tagged as observed, inferred or implication, and a feature not seen on the pages checked is written as not observed, never as absent.
What You Get
- A dated profile per competitor from one template
- Raw data saved by competitor and date: scrapes, SEO JSON and reviews
- Sections on positioning, product, pricing, customers, SEO and strengths and weaknesses
- Competitive implications written as questions or options
- A cross-competitor summary file
Common Use Cases
- Profiling three rivals before a pricing change
- Checking what a competitor charges and claims today
- Updating an old profile and noting what changed
- Testing whether a conclusion about a rival is fair
What Makes It Unique
It separates observed, inferred and implied claims, forbids stating a rival’s motive, and treats fetched competitor pages as data to analyze, ignoring any instructions hidden in them. Metrics with no data source are marked unavailable.
Under The Hood
The SKILL.md and two references (a template file and a tool reference) are loaded as context. It tells the agent to scrape competitor pages with Firecrawl or WebFetch, pull SEO data through DataForSEO or another connected MCP, and write files under a competitor-profiles folder in your project. It ships no scripts; the tools need your own accounts.