saas-strategy

Overview

This skill delivers McKinsey/Deloitte-style strategic analysis of SaaS companies. It activates when a user requests competitive intelligence, market positioning assessment, product opportunity mapping, or AI readiness evaluation. The main entry point is the /saas-analyze command.

The skill runs a structured multi-agent workflow — beginning with an iterative Discovery Call to scope the engagement, followed by parallel research, synthesis, source verification, and final report generation.

Entry Point

The /saas-analyze command launches the full workflow. Users may invoke it with or without arguments:

• /saas-analyze — starts the Discovery Call intake with no prior context
• /saas-analyze [company name] — pre-fills the subject; Discovery Call continues from Step A2
• /saas-analyze [company name] [strategic question] — pre-fills subject and question; Discovery Call continues from Step A3

Agent Roster

Six specialized agents execute the analysis. Each has a defined role and a bounded scope:

Agent	Role	Phase
Discovery Facilitator	Conducts the structured intake interview, builds the Analysis Brief, confirms scope	A
Market Intelligence Analyst	Researches market size, trends, TAM/SAM/SOM, industry reports, technology adoption curves	B
Competitive Intelligence Analyst	Profiles competitors, maps features, pricing, positioning, job postings, and recent moves	B
Product & AI Analyst	Audits product capabilities, reviews GitHub/changelog, scores AI maturity, identifies gaps	B
Lead Strategist	Synthesizes all research streams into strategic insights, recommendations, and opportunity rankings	D
Source Verifier	Reviews all claims across the draft report, annotates confidence levels, flags anything below the 20% verified-data threshold	E
Report Architect	Assembles the final deliverables: five Markdown files plus a self-contained HTML report with inline SVG charts	F

Discovery Call Intake Process

The Discovery Call (Phase A) is iterative — the Discovery Facilitator asks questions one group at a time, waits for user responses, and refines the scope before proceeding. It never fires all questions at once.

The five steps are:

1. Subject Identification — confirm the company name, website, and market
2. Output Language — set the language for all deliverables (default: English)
3. Strategic Question — establish what the user actually wants to answer (e.g., "Where should we invest product resources in the next 12 months?")
4. Data Source Inventory — collect any internal data the user can provide (codebase URLs, MRR, churn, roadmap, G2 profile, etc.)
5. Workspace Setup & Scope Confirmation — generate analysis-brief.md, confirm what can be analyzed with high vs. low confidence given available data

Only after Step 5 is confirmed does the workflow advance to Phase B.

Reference Files

Agents should consult the following references throughout execution:

• references/strategic-frameworks.md — descriptions of all analytical frameworks used, when to apply them, and which agent owns each
• references/analysis-phases.md — detailed phase-by-phase workflow including timing estimates, re-research triggers, and language propagation rules
• references/output-formats.md — exact structure for all Markdown outputs, HTML report sections, and SVG chart specifications with coordinates
• references/data-sources.md — ranked list of trusted public data sources by category, with guidance on what to extract and what confidence level to assign
• references/citation-protocol.md — citation format, four confidence levels (VERIFIED / ESTIMATED / INFERRED / UNVERIFIED), and the anti-hallucination checklist every agent must run before writing a number into a report