feat(seo-engine): implement competitor scraper, MDX draft editor, and strategy report generator

packages/seo-engine/src/steps/content-gap.ts

@@ -0,0 +1,84 @@
+import { llmJsonRequest } from "../llm-client.js";
+import type { TopicCluster } from "../types.js";
+
+export interface ExistingPage {
+  url: string;
+  title: string;
+}
+
+export interface ContentGap {
+  recommendedTitle: string;
+  targetKeyword: string;
+  relatedCluster: string;
+  priority: "high" | "medium" | "low";
+  rationale: string;
+}
+
+const CONTENT_GAP_SYSTEM_PROMPT = `
+You are a senior SEO Content Strategist. Your job is to compare a set of TOPIC CLUSTERS
+(keywords the company should rank for) against the EXISTING PAGES on their website.
+
+### OBJECTIVE:
+Identify content gaps — topics/keywords that have NO corresponding page yet.
+For each gap, recommend a page title, the primary target keyword, which cluster it belongs to,
+and a priority (high/medium/low) based on commercial intent and relevance.
+
+### RULES:
+- Only recommend gaps for topics that are genuinely MISSING from the existing pages.
+- Do NOT recommend pages that already exist (even if the title is slightly different — use semantic matching).
+- Priority "high" = commercial/transactional intent, directly drives revenue.
+- Priority "medium" = informational with strong industry relevance.
+- Priority "low" = broad, top-of-funnel topics.
+- LANGUAGE: Match the language of the project context (if German context, recommend German titles).
+
+### OUTPUT FORMAT:
+{
+  "contentGaps": [
+    {
+      "recommendedTitle": "string",
+      "targetKeyword": "string",
+      "relatedCluster": "string",
+      "priority": "high" | "medium" | "low",
+      "rationale": "string"
+    }
+  ]
+}
+`;
+
+export async function analyzeContentGaps(
+  topicClusters: TopicCluster[],
+  existingPages: ExistingPage[],
+  config: { openRouterApiKey: string; model?: string },
+): Promise<ContentGap[]> {
+  if (topicClusters.length === 0) return [];
+  if (existingPages.length === 0) {
+    console.log(
+      "[Content Gap] No existing pages provided, skipping gap analysis.",
+    );
+    return [];
+  }
+
+  const userPrompt = `
+TOPIC CLUSTERS (what the company SHOULD rank for):
+${JSON.stringify(topicClusters, null, 2)}
+
+EXISTING PAGES ON THE WEBSITE:
+${existingPages.map((p, i) => `${i + 1}. "${p.title}" — ${p.url}`).join("\n")}
+
+Identify ALL content gaps. Be thorough but precise.
+`;
+
+  try {
+    const { data } = await llmJsonRequest<{ contentGaps: ContentGap[] }>({
+      model: config.model || "google/gemini-2.5-pro",
+      apiKey: config.openRouterApiKey,
+      systemPrompt: CONTENT_GAP_SYSTEM_PROMPT,
+      userPrompt,
+    });
+
+    return data.contentGaps || [];
+  } catch (err) {
+    console.error("[Content Gap] Analysis failed:", (err as Error).message);
+    return [];
+  }
+}