market-intel

Router skill. Resolves one topic, then Reads EXACTLY ONE playbook from routes:. Never loads the whole cluster — one topic per invocation. Covers market/competitive intelligence: positioning, pricing, sector context, benchmarks, partnerships. NOT product specs, GTM execution, or financial modeling — route those elsewhere. [INFERENCIA]

Routing

• topic (required): one enum value (params.topic.enum). Infer from the request; ask only if genuinely ambiguous between two topics. [CONFIG]
• depth (default quick): deep → apply the playbook exhaustively with verification at each step; quick → essentials only. [CONFIG]
• Read that topic's file from routes: — nothing else — then run its spine: Discover → Analyze → Execute → Validate. [DOC]

Topic cues: competitor matrix/stack/SWOT → competitive-intelligence; where-we-win → competitive-positioning; vs-peer metrics → benchmarking-analysis; TAM/trends/demand → market-intelligence; vertical/regulatory → sector-intelligence; messaging/channel → marketing-context; ally/OEM → partnership-strategy; packaging/tiers → pricing-strategy. [INFERENCIA]

Acceptance criteria

• Exactly one playbook loaded; topic ∈ enum. [DOC]
• Every non-obvious claim tagged ([CÓDIGO] [CONFIG] [DOC] [INFERENCIA] [SUPUESTO]); one tag per claim, consistent throughout. [DOC]
• No invented prices — FTE-months + disclaimers only. [DOC]
• depth=deep runs the playbook's validation step before output. [DOC]

Anti-patterns

• Loading 2+ topic files "to be safe" — defeats the router. [INFERENCIA]
• Guessing an ambiguous topic instead of asking. [INFERENCIA]
• Emitting market figures as fact without a {WEB}+citation or [DOC] tag. [DOC]

Quality gates: constitution v6.0.0 (enforcement), evidence tags, script-first rule. The guardian applies assets/quality-rubric.json and assets/routing-checklist.md before any deliverable is emitted. [CONFIG]