@maintainabilityai/research-runner 0.1.11 → 0.1.12

package/dist/runner/nodes/dedupe-and-rank.js

@@ -30,6 +30,19 @@ function canonicalizeUrl(rawUrl) {
         return rawUrl.trim().toLowerCase();
     }
 }
+/**
+ * Per-provider quota for the top-N output. Without these floors, Tavily
+ * (normalized scores 0.9–1.0) crushes every other provider in pure
+ * global ranking — synth would see zero HN signal and zero patent
+ * coverage. Quotas sum to topN's default (20). Any unused slack
+ * spills over to the highest-scoring non-quota entries across providers.
+ */
+const PROVIDER_QUOTA = {
+    tavily: 8,
+    arxiv: 5,
+    uspto: 4,
+    hackernews: 3,
+};
 function dedupeAndRank(opts) {
     const topN = opts.topN ?? 20;
     const retrievedAt = opts.retrievedAt ?? new Date().toISOString();
@@ -72,12 +85,46 @@ function dedupeAndRank(opts) {
             });
         }
     }
-    const ranked = [...bucket.values()]
-        .map(a => {
+    const allEntries = [...bucket.values()].map(a => {
         const recall = 1 + 0.15 * (a.queries.size - 1);
         const composite = Math.min(1, a.scoreSum * recall / Math.max(1, a.occurrences));
         return { aggregated: a, composite };
-    })
+    });
+    // Phase 1 — per-provider quota: take each provider's top-K (K from PROVIDER_QUOTA).
+    // Phase 2 — spillover: fill the remaining budget with the next-highest entries
+    //          from anywhere, including providers that have already filled their quota.
+    // Phase 3 — re-sort the combined set by composite score for stable display order.
+    const used = new Set();
+    const picks = [];
+    for (const provider of Object.keys(PROVIDER_QUOTA)) {
+        const k = PROVIDER_QUOTA[provider];
+        if (k === 0) {
+            continue;
+        }
+        const fromProvider = allEntries
+            .filter(e => e.aggregated.provider === provider)
+            .sort((a, b) => b.composite - a.composite)
+            .slice(0, k);
+        for (const e of fromProvider) {
+            if (used.has(e.aggregated.canonicalUrl)) {
+                continue;
+            }
+            picks.push(e);
+            used.add(e.aggregated.canonicalUrl);
+        }
+    }
+    const remainingBudget = Math.max(0, topN - picks.length);
+    if (remainingBudget > 0) {
+        const spillover = allEntries
+            .filter(e => !used.has(e.aggregated.canonicalUrl))
+            .sort((a, b) => b.composite - a.composite)
+            .slice(0, remainingBudget);
+        for (const e of spillover) {
+            picks.push(e);
+            used.add(e.aggregated.canonicalUrl);
+        }
+    }
+    const ranked = picks
         .sort((a, b) => b.composite - a.composite)
         .slice(0, topN);
     return ranked.map((entry, i) => ({

package/dist/runner/nodes/format-for-human.js

@@ -46,7 +46,11 @@ function providerSection(label, emoji, provider, sources, totalCount) {
     for (const s of sources) {
         const authors = s.authors && s.authors.length > 0 ? ` — _${s.authors.slice(0, 3).join(', ')}${s.authors.length > 3 ? ' et al.' : ''}_` : '';
         const date = s.published_at ? ` _(${s.published_at.slice(0, 10)})_` : '';
-        lines.push(`- **[\`${s.id}\`] [${s.title}](${s.url})** — score ${s.salience_score.toFixed(2)}${date}${authors}`);
+        // Render the citation id as standalone inline code so the synth
+        // agent (and a downstream PRD agent) can grep `\bS\d+\b` cleanly.
+        // The earlier form **[`S1`] [Title](url)** broke GitHub's markdown
+        // parser (it tried to interpret the brackets as a reference link).
+        lines.push(`- \`${s.id}\` **[${s.title}](${s.url})** — score ${s.salience_score.toFixed(2)}${date}${authors}`);
         lines.push(`  > ${shortExcerpt(s.excerpt)}`);
     }
     lines.push('');
@@ -159,6 +163,6 @@ function formatForHuman(opts) {
     lines.push('');
     lines.push('---');
     lines.push('');
-    lines.push(`🤖 Generated by \`research-runner archeologist\` v${'pkg'}. Run id: \`${runId}\`.`);
+    lines.push(`🤖 Generated by \`research-runner archeologist\`. Run id: \`${runId}\` (see the Hatter's Tag for agent version + audit chain).`);
     return { body: lines.join('\n') };
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@maintainabilityai/research-runner",
-  "version": "0.1.11",
+  "version": "0.1.12",
   "description": "Research + PRD agent runner — orchestrates the Archeologist and PRD pipelines for the MaintainabilityAI governance mesh",
   "license": "MIT",
   "author": "MaintainabilityAI",