business-analysis-generator

Business Analysis Generator

A Claude Code skill that runs a consulting-grade business analysis of a specific company and turns it into a reusable, cross-system prompt library. Built for AI consulting audits, client onboarding, and implementation planning.

What it does

The skill does two things: it analyzes a company the way a consulting engagement would, then it converts that analysis into prompts the business can run itself. The prompts are where the leverage lands. Most "AI prompt" advice produces prompts that a single tool could already answer on its own. This skill produces the prompts actually worth having: ones that pull data from two or more systems (CRM, accounting, email, calendar, project tooling) to answer a real business question. The leverage of an AI prompt shows up when it does work that crosses system boundaries.

Given a business, the skill works like a consulting engagement:

1. Research first. A delegated research subagent does the homework a professional consulting firm's analyst would do before kickoff: a cited company fact base, an industry and market analysis (market sizing, value chain, Porter's Five Forces, competitive landscape, regulatory context, benchmarks), a SWOT, and a ranked set of hypotheses about where the money, risk, and friction sit. Depth scales to the company's footprint, and every claim is sourced.
2. Adaptive discovery. The research generates the interview, not the other way around. Questions are tailored to the specific business and ordered by the hypotheses they test, so they vary with what the research found. The six standard clusters (business model, systems, leadership pain, data quality, constraints, operating preferences) act as a coverage checklist.
3. Generation. Produces a prioritized library of 4 to 10 prompts, each with a full five-part specification, after restating its understanding and getting explicit confirmation.

The two outputs run at different speeds. The analysis is built once and refreshed only occasionally, or when something material changes. The prompt library is what the business runs on a recurring cadence, often monthly, to get fresh analysis from live data without redoing the research each time.

The five-part spec

Every prompt in the output is fully specified, not just listed:

• Connectors required. The actual tools and data sources, named to the business's real stack.
• Grounding decisions. The judgment calls (thresholds, comparison cohorts, time windows) that make the prompt produce meaningful results.
• Good response. What a useful output looks like, concrete enough to evaluate against.
• Bad response. The failure modes that separate a working prompt from a generic one.
• Industry adaptation. How the prompt is tuned for the specific business and its regulatory context.

Installing

This is a Claude Code skill. Clone it into your skills directory:

git clone https://github.com/adamhobb11/business-analysis-generator.git \
  ~/.claude/skills/business-analysis-generator

Claude Code picks it up automatically. Skills live in ~/.claude/skills/<name>/ with a SKILL.md at the root.

Using it

In Claude Code, trigger the skill by asking for a business analysis or a prompt library for a business. For example:

• "Run a business analysis for a 12-person accounting firm and turn it into prompts."
• "Build an AI readiness audit for my HVAC distribution business."
• "What cross-system prompts would work for a dental practice?"

The skill starts the discovery interview first. It does not generate prompts until you confirm it has understood the business correctly.

What's inside

SKILL.md holds the full skill: the research process, the adaptive discovery process, the generation rules, and a seed library of 13 proven prompt archetypes spanning revenue and pipeline, cash and margin, retention signals, operations, and industry-specific use cases.

License

MIT. See LICENSE.