diff --git a/server/scripts/migrate-json-to-pg.ts b/server/scripts/migrate-json-to-pg.ts
new file mode 100644
index 0000000..a84b516
--- /dev/null
+++ b/server/scripts/migrate-json-to-pg.ts
@@ -0,0 +1,106 @@
+/**
+ * One-time migration: import existing igdb-cache.json and igdb-metadata.json
+ * into PostgreSQL. Run with: DATABASE_URL=... bun server/scripts/migrate-json-to-pg.ts
+ */
+import { readFileSync } from "node:fs"
+import { dirname, join } from "node:path"
+import { fileURLToPath } from "node:url"
+import postgres from "postgres"
+
+const __dirname = dirname(fileURLToPath(import.meta.url))
+const DATA_DIR = join(__dirname, "../src/data")
+
+const DATABASE_URL = process.env.DATABASE_URL
+if (!DATABASE_URL) {
+	console.error("DATABASE_URL is required")
+	process.exit(1)
+}
+
+const sql = postgres(DATABASE_URL)
+
+const BATCH_SIZE = 500
+
+async function migrateResolutions() {
+	const file = join(DATA_DIR, "igdb-cache.json")
+	let data: Record<string, { igdbId: number | null }>
+	try {
+		data = JSON.parse(readFileSync(file, "utf-8"))
+	} catch {
+		console.log("[migrate] No igdb-cache.json found, skipping resolutions")
+		return
+	}
+
+	const entries = Object.entries(data)
+	console.log(`[migrate] Importing ${entries.length} resolution entries...`)
+
+	for (let i = 0; i < entries.length; i += BATCH_SIZE) {
+		const batch = entries.slice(i, i + BATCH_SIZE)
+		const values = batch.map(([key, val]) => {
+			const [source, ...rest] = key.split(":")
+			const sourceId = rest.join(":")
+			return { cache_key: key, source, source_id: sourceId, igdb_id: val.igdbId }
+		})
+
+		await sql`
+			INSERT INTO igdb_resolutions ${sql(values, "cache_key", "source", "source_id", "igdb_id")}
+			ON CONFLICT (cache_key) DO NOTHING
+		`
+
+		console.log(
+			`[migrate] Resolutions: ${Math.min(i + BATCH_SIZE, entries.length)}/${entries.length}`,
+		)
+	}
+}
+
+async function migrateMetadata() {
+	const file = join(DATA_DIR, "igdb-metadata.json")
+	let data: Record<
+		string,
+		{
+			summary: string | null
+			coverImageId: string | null
+			screenshots: string[]
+			videoIds: string[]
+			genres: string[]
+			aggregatedRating: number | null
+			releaseDate: string | null
+			developers: string[]
+		}
+	>
+	try {
+		data = JSON.parse(readFileSync(file, "utf-8"))
+	} catch {
+		console.log("[migrate] No igdb-metadata.json found, skipping metadata")
+		return
+	}
+
+	const entries = Object.entries(data)
+	console.log(`[migrate] Importing ${entries.length} metadata entries...`)
+
+	for (let i = 0; i < entries.length; i += BATCH_SIZE) {
+		const batch = entries.slice(i, i + BATCH_SIZE)
+		const values = batch.map(([canonicalId, meta]) => ({
+			canonical_id: canonicalId,
+			summary: meta.summary,
+			cover_image_id: meta.coverImageId,
+			screenshots: JSON.stringify(meta.screenshots),
+			video_ids: JSON.stringify(meta.videoIds),
+			genres: JSON.stringify(meta.genres),
+			aggregated_rating: meta.aggregatedRating,
+			release_date: meta.releaseDate,
+			developers: JSON.stringify(meta.developers),
+		}))
+
+		await sql`
+			INSERT INTO igdb_metadata ${sql(values, "canonical_id", "summary", "cover_image_id", "screenshots", "video_ids", "genres", "aggregated_rating", "release_date", "developers")}
+			ON CONFLICT (canonical_id) DO NOTHING
+		`
+
+		console.log(`[migrate] Metadata: ${Math.min(i + BATCH_SIZE, entries.length)}/${entries.length}`)
+	}
+}
+
+await migrateResolutions()
+await migrateMetadata()
+console.log("[migrate] Done!")
+await sql.end()