analyze-competition - SKILL.md Agent Skill

name: analyze-competition description: Competitive analysis — identify competitors, build feature matrix, assess positioning, and produce go/no-go verdict with kill-criteria framing. Use when "competitive analysis", "compare competitors", "positioning vs", "competitive landscape", "battlecard", "who are the competitors". Outputs analysis/competitive-analysis.md. allowed-tools: ["Read", "Write", "Glob", "Grep"] model: sonnet context: subagent user-invocable: true argument-hint: ""

Analyze Competition

Produce a competitive analysis that scales to available input. Thin input → scout report. Rich input → deep matrix.

Inputs

Read available context (any order): PRD, market notes, prior analysis, pasted Perplexity output, competitor URLs. If $ARGUMENTS names a product/market, use that as seed. If nothing found, ask user for product name + target segment.

Methodology

Discover — direct, indirect, adjacent, substitute competitors. Use 3-tier search: known names → category search → adjacent categories.
Compare — side-by-side feature/capability matrix (✅/❌/🟡).
Position — pricing/business model, target segment, UVP, moat per competitor.
Assess — kill-criteria check against thresholds. Be honest about where target product loses.
Enrich — always append a tailored Perplexity deep-search prompt for gaps found.

Output

Write analysis/competitive-analysis.md using the template in references/methodology.md § Output Template.

Core sections (always present): Competitors table, Kill-Criteria Check, Sources, Perplexity prompt. Optional sections (when data supports): Feature Matrix, Positioning, Gaps & Opportunities. Mark missing sections [insufficient data — enrich via Perplexity prompt below]. Optional frameworks (apply only when trigger condition met): Market Timing, Trajectory, ERRC, Entry Barriers. See references/methodology.md § Optional Frameworks.

Key Rules

Cite sources for every factual claim
Adapt sources to industry (see references/methodology.md § Industry Sources)
Kill-criteria: if any criterion breaches threshold → explicit kill recommendation
No speculation — mark uncertain data with confidence level (high/medium/low)
See references/methodology.md for output template, Perplexity prompt, and FR sources
Next step: At end of output, suggest /ux-strategize empathy-map {subject} to start building UX artifacts from the analysis