brand-coherence-auditlisted

Audit the discursive coherence of a brand or person across the web (owned site + off-site sources), in FR or EN. Extract structured claims (positioning, offer, expertise, proofs) from up to 15 sources, build a cross-source coherence matrix, surface inconsistencies that hurt LLM citation likelihood (Generative Engine Optimization), and deliver a prioritized markdown report with sub-scores per axis and a weighted aggregate. Use when the user wants to check whether their brand or personal narrative is consistent enough for AI search engines (Perplexity, ChatGPT Search, Google AI Mode, Brave Ask) to cite them confidently. Trigger for requests such as "audit la cohérence du discours de [marque/personne]", "est-ce que mon discours est cohérent sur le web", "check ma cohérence narrative pour les LLM", "brand coherence audit", "discourse consistency audit", or "verify my brand narrative across sources".
JuJu78/brand-coherence-audit · ★ 0 · AI & Automation · score 70

Install: claude install-skill JuJu78/brand-coherence-audit

# Brand Discourse Coherence Audit Use this skill to measure whether a **brand** or a **person** tells a consistent story across the web. The unit of analysis is the **structured claim** (positioning, offer, expertise, proof), not the keyword and not the sentiment. Output is a typed cross-source coherence matrix and a prioritized list of inconsistencies that risk hurting Generative Engine Optimization (GEO). This skill exists because LLMs running AI search (Perplexity, ChatGPT Search, Google AI Mode, Brave Ask, Claude with web) perform **entity reconciliation** before citing a source. When a brand says one thing on its website, another on LinkedIn, and a third in a podcast, the LLM lowers its citation confidence and falls back to safer sources. This skill surfaces those gaps before they cost visibility. This skill is **distinct from** `entity-territory-audit` and `semantic-gap-blindspot-auditor`. Those analyze one URL against competitors on a topic. This one analyzes one **brand or person** across **its own multi-source footprint**, regardless of competitors. ## Required Inputs - `subject` - the brand or person name being audited (e.g. "Yann LeCun", "Stripe", "Anthropic") ## Optional Inputs - `type` - `person` | `brand`. Auto-detected from a Brave search if omitted. - `depth` - `quick` (5 sources, fast) | `standard` (8-10 sources, default) | `deep` (up to 15 sources, includes podcast/video transcripts). - `sources` - explicit list of URLs to audit. When provided, skip d