What’s inside

One SKILL.md (the playbook) + one starter template. A preview:

Competitor Researcher

When to use

• You want a Monday-morning snapshot of what competitors changed last week: pricing, features, announcements, hiring
• Job listings are a leading indicator in your market and you want someone reading the signals for you
• You need a one-page competitive brief before a strategy or product meeting
• Pricing pages shift often and you want changes flagged with screenshots as proof
• You'd rather read takeaways than browse five competitors' sites yourself

Workflow

Name 3–7 competitors and what to watch on each site: pricing page, changelog, announcements, careers page. Be specific per site — "keep an eye on Acme" is not a spec.

1. Set the watch list (input: competitor names → output: watch list + signals)

On your chosen cadence (the classic is Monday 8am), visit each site, record starting prices and core features, and capture homepage screenshots. Read-only: never submit forms, never apply for jobs, never create accounts.

1. Fetch on schedule (input: watch list → output: snapshots + screenshots)

Compare and list only what changed. No change means no entry. This step is what makes the brief worth reading — a dump of everything is not intelligence.

1. Diff against last time (input: this week's snapshots + last week's → output: change list)

Read new job descriptions and infer what each hire implies about product direction. Label every inference as inference, not fact — a JD is a signal, not an announcement.

1. Read the hiring signals (input: new job postings → output: inferred intentions)

One page: changes summary, the single most important item, and three takeaways for your team. Every claim links to a screenshot or page with a capture date. Anything inferred is labeled as such.

1. Write the one-page brief (input: change list + signals → output: brief)

Output spec

• One-page brief per run: changes since last run, the top item, three takeaways
• Evidence appendix: screenshots, URLs, capture dates
• A stored snapshot archive so the next run can diff against this one
• Companion template.md skeleton — fill in the blanks and use it

Good vs. bad examples

• ❌ Bad signal: "Competitor hired a PM — they must be launching something big soon." (stated as fact)
• ✅ Good signal: "Competitor posted 3 ML-engineer roles (links, captured 2026-10-05) — *inference*: likely investing in ML features; *fact*: they have announced nothing" — inference labeled, evidence linked
• ❌ Bad brief: a ten-page dump of everything on competitors' sites, no changes highlighted, no takeaways
• ✅ Good brief: "Changed: Acme raised starter price $19→$24 (screenshot 2026-10-13) — first pricing change in 18 months. Takeaway 1: test our own price-sensitivity messaging this quarter…"
• ❌ Bad research: age

Download the full skill pack (ZIP)