Dineway Content Competition

Dineway Content Competition

Treat crawl and SERP retrieval as evidence acquisition; perform gap diagnosis, clustering, prioritization, and recommendations locally in the Agent. Return one canonical report and typed Competition payload to the active Pipeline Stage.

Inputs and acquisition

• Target URL or planned topic, locale, market, industry context, site inventory, and accepted Research Result when present.
• Current SERP observations and up to three direct comparison URLs for a focused page analysis.
• Local HTTP crawl and Browser Use first. Use Forgeway DataForSEO/Firecrawl only for managed-key or infrastructure-heavy live SERP/page acquisition. Never require the user to configure provider keys when Dineway owns the connector.

Record URL, title, domain, rank, word count, headings, internal/external links, collected time, availability, and raw artifact reference. Filter irrelevant directories, login pages, social noise, copied pages, and non-comparable intent before analysis.

Workflow

1. Confirm target intent and comparison set. A ranking page with different intent is evidence, not a valid template.
2. Compare topic coverage, section depth, heading structure, questions, tables, media, citations, entity clarity, freshness, and links.
3. Separate strengths from gaps. Link every conclusion to observation IDs; do not call crawler output a semantic fact.
4. Calculate an opportunity score only when all four factor inputs are measured: impact 35%, inverse effort 25%, speed 20%, strategic value 20%. Otherwise retain missing factors and score as unmeasured under the active Pipeline scoring policy.
5. Cluster related keywords/topics locally. Each cluster records members and rationale. Use dineway-seo-cluster if the recommendation becomes a formal Dineway content architecture.
6. Recommend differentiated scope, not imitation. Identify quick wins, growth work, and strategic work as views over measured evidence rather than separate authorities.

Stage completion contract

Write jobs/<job-id>/competition/report.json, then return its canonical content and a typed payload containing target URL, manualCompetitorUrls, location and industry context, opportunity score, competitors with title/rank/strengths/gaps/observation IDs, local clusters, and recommendations. Preserve the source analysis phases as bounded artifact references: pageCrawlArtifactRef, rankingQueriesArtifactRef, serpCompetitorsArtifactRef, competitorAnalysisArtifactRef, and gapAnalysisArtifactRef; use null when a phase is unavailable. Include source timestamps, all observation IDs, provenance, and any bounded raw artifact reference.

Return to dineway-content-pipeline, which calls content_pipeline_stage_complete to derive the receipt and create and accept the immutable Competition Result. Do not call granular Result operations or create a Brief, Draft, cluster, or CMS change from this Skill without a separate Stage and owning Skill.