agentleFS
Kikobazz123Kikobazz123/lordgen-competition-demo/.claude/skills/business-research/SKILL.md

business-research

Skill0 starsChanged 3 months agoMITView on GitHub

Research a target business using public web sources and produce structured, evidence-graded findings — verified facts, inference, and confidence kept separate. Use when researching the locked LordGen demo business or a candidate business.

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use Claude Code's built-in WebSearch and WebFetch tools only — no external research MCP server (per the approved architecture, …

    line 14
  2. Very specific:

    If firecrawl also fails (e.g.

    line 22
  3. Distinctive:

    Never fabricate page content.

    line 23
  4. Very specific:

    Write data/research.json with this structure:

    line 29
  5. Very specific:

    Every fact and pain_point must carry a source_url (or "source_url": null with an explanation if genuinely unsourceable) and a type of …

    line 50
  6. Distinctive:

    If a source is unreachable, record that as a data caveat rather than silently dropping it or guessing at its content.

    line 52

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Business Research

Purpose

Research a target business and produce structured findings that a human can trust: what's actually known, what's a reasoned guess, and how confident each claim is. Never present inference as verified fact (CLAUDE.md §11).

Method

Use Claude Code's built-in WebSearch and WebFetch tools only — no external research MCP server (per the approved architecture, docs/architecture.md). Prioritize, in order:

  1. The business's own website — services, hours, service area, stated policies/claims.
  2. Regulatory/accreditation sources (e.g. BBB) — rating, accreditation status, complaint patterns.
  3. Independent review aggregators (Google, Yelp, Trustpilot, industry-specific sites) — ratings, review counts, recurring themes.

Some sources (Yelp, Trustpilot, and similar) commonly block automated WebFetch access (HTTP 403). When that happens, in order:

  1. Retry with firecrawl_scrape (proxy: "stealth" or "enhanced") — approved 2026-08-13 specifically for this fallback case (docs/architecture.md). This gets past most bot-blocks and returns primary-source content, not a summary.
  2. If firecrawl also fails (e.g. still 403, or the content is behind client-side JS pagination that doesn't render), fall back to WebSearch result summaries as a last resort.
  3. Never fabricate page content.

Mark facts sourced via WebSearch summary (rather than a direct page read, by either tool) with "confidence": "low" and a "caveat" explaining why. Facts read directly (via WebFetch or firecrawl_scrape) can carry "confidence": "high" for the fact's existence/text — but a single review/account being real and directly-read is not the same as it being representative; that judgment still belongs in confidence on the pain_points/automation_opportunities entries that cite it, not in inflating the underlying fact.

Required Output

Write data/research.json with this structure:

{
  "business": { "name": "", "category": "", "founded": "", "address": "", "phone": "", "service_area": "", "hours": "" },
  "research_date": "YYYY-MM-DD",
  "facts": [
    { "fact": "", "type": "verified|inference", "source": "", "source_url": "", "confidence": "high|medium|low", "caveat": "" }
  ],
  "pain_points": [
    { "description": "", "type": "verified|inference", "evidence": "", "confidence": "high|medium|low", "supporting_fact_refs": [] }
  ],
  "automation_opportunities": [
    { "opportunity": "", "rationale": "", "confidence": "high|medium|low", "supporting_fact_refs": [] }
  ],
  "data_caveats": [""],
  "overall_confidence": "high|medium|low"
}

Rules:

  • Every fact and pain_point must carry a source_url (or "source_url": null with an explanation if genuinely unsourceable) and a type of verified or inference — never omit type.
  • automation_opportunities are candidate opportunities only. Scoring/ranking them is the opportunity-score skill's job (Phase 3), not this one.
  • If a source is unreachable, record that as a data caveat rather than silently dropping it or guessing at its content.
  • overall_confidence reflects the weakest link, not the average — e.g. if the core pain-point evidence is only "low," overall confidence should not read "high."

Out of Scope

This skill only researches and writes data/research.json. It does not score opportunities, write proposals, or take any external action.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.