@ainyc/canonry 4.109.0 → 4.111.0

package/dist/{chunk-3HKBUYIY.js → chunk-CVXBE3ST.js}

@@ -270,7 +270,7 @@ import {
   wordpressSchemaDeployResultDtoSchema,
   wordpressSchemaStatusResultDtoSchema,
   wordpressStatusDtoSchema
-} from "./chunk-WXAHSQOX.js";
+} from "./chunk-ZT7GVLS3.js";
 
 // src/intelligence-service.ts
 import { eq as eq38, desc as desc18, asc as asc6, and as and28, ne as ne5, or as or6, inArray as inArray14, gte as gte7, lte as lte4 } from "drizzle-orm";
@@ -1160,6 +1160,14 @@ var discoverySessions = sqliteTable("discovery_sessions", {
   dedupClusterMinSims: text("dedup_cluster_min_sims", { mode: "json" }).$type(),
   dedupBandPairFraction: real("dedup_band_pair_fraction"),
   dedupPairsTotal: integer("dedup_pairs_total"),
+  // Seed provider set (canonical order; null = legacy / Gemini-only) — part of
+  // session identity for consolidation — and per-provider candidate counts.
+  seedProviders: text("seed_providers", { mode: "json" }).$type(),
+  seedProviderCounts: text("seed_provider_counts", { mode: "json" }).$type(),
+  // TRUE canonical count after dedup, BEFORE the probe-budget slice. seed_count
+  // is post-truncation (probedCanonicals.length), so a deliberately small
+  // maxProbes deflates it; this column is the honest numerator.
+  canonicalCount: integer("canonical_count"),
   dedupThreshold: real("dedup_threshold"),
   probeCount: integer("probe_count"),
   citedCount: integer("cited_count"),
@@ -3452,6 +3460,25 @@ var MIGRATION_VERSIONS = [
       `ALTER TABLE discovery_sessions ADD COLUMN dedup_band_pair_fraction REAL`,
       `ALTER TABLE discovery_sessions ADD COLUMN dedup_pairs_total INTEGER`
     ]
+  },
+  {
+    version: 93,
+    name: "discovery-session-seed-providers",
+    statements: [
+      // Seed provider set (consolidation identity) + per-provider candidate
+      // counts. Additive + nullable — legacy rows stay null (downgrade-safe).
+      `ALTER TABLE discovery_sessions ADD COLUMN seed_providers TEXT`,
+      `ALTER TABLE discovery_sessions ADD COLUMN seed_provider_counts TEXT`
+    ]
+  },
+  {
+    version: 94,
+    name: "discovery-session-canonical-count",
+    statements: [
+      // True post-dedup, pre-truncation canonical count (seed_count is
+      // post-truncation). Additive + nullable (downgrade-safe).
+      `ALTER TABLE discovery_sessions ADD COLUMN canonical_count INTEGER`
+    ]
   }
 ];
 function rebuildBacklinkTableWithSource(tx, table) {
@@ -18119,6 +18146,7 @@ var routeCatalog = [
             properties: {
               icpDescription: { type: "string", description: "Free-text ICP. Required if the project does not have spec.icpDescription stored." },
               buyerDescription: { type: "string", description: "Who evaluates or buys the offering, separate from the ICP. When present, every generated seed query is anchored on this buyer and the seed prompt enforces buyer-fit." },
+              seedProviders: { type: "array", items: { type: "string", enum: ["gemini", "openai"] }, minItems: 1, description: "Which providers generate seed candidates. Omit for the Gemini-only default. Canonicalized (deduped + sorted); part of the session consolidation identity." },
               dedupThreshold: { type: "number", description: "Cosine similarity threshold for clustering. Defaults to 0.95." },
               maxProbes: { type: "integer", description: "Max canonical queries to probe in this session. Default 100, hard cap 500." },
               probeConcurrency: {
@@ -33812,6 +33840,8 @@ async function discoveryRoutes(app, opts) {
         "icpDescription is required. Pass it in the request body or store it on the project (spec.icpDescription)."
       );
     }
+    const requestedSeedProviders = parsed.data.seedProviders ?? null;
+    const seedProviders = requestedSeedProviders && !(requestedSeedProviders.length === 1 && requestedSeedProviders[0] === "gemini") ? requestedSeedProviders : null;
     const locations = orderLocationsDefaultFirst(
       resolveLocations(project.locations, parsed.data.locations),
       project.defaultLocation
@@ -33841,6 +33871,10 @@ async function discoveryRoutes(app, opts) {
         // never reuses — a one-time, bounded (2h window) non-reuse after
         // upgrade, never a wrong reuse.
         locations.length === 0 ? or5(isNull(discoverySessions.locations), eq35(discoverySessions.locations, locations)) : eq35(discoverySessions.locations, locations),
+        // Seed provider set is identity: a different phrasing distribution
+        // must never reuse another set's session. Null (= Gemini-only
+        // default, including explicit ['gemini']) matches legacy rows.
+        seedProviders == null ? isNull(discoverySessions.seedProviders) : eq35(discoverySessions.seedProviders, seedProviders),
         inArray12(discoverySessions.status, [
           DiscoverySessionStatuses.queued,
           DiscoverySessionStatuses.seeding,
@@ -33860,6 +33894,7 @@ async function discoveryRoutes(app, opts) {
         status: DiscoverySessionStatuses.queued,
         icpDescription,
         buyerDescription: parsed.data.buyerDescription ?? null,
+        seedProviders,
         locations,
         dedupThreshold: parsed.data.dedupThreshold,
         competitorMap: [],
@@ -33896,6 +33931,7 @@ async function discoveryRoutes(app, opts) {
       projectId: project.id,
       icpDescription,
       buyerDescription: parsed.data.buyerDescription,
+      seedProviders: seedProviders ?? void 0,
       dedupThreshold: parsed.data.dedupThreshold,
       maxProbes: parsed.data.maxProbes,
       probeConcurrency: parsed.data.probeConcurrency,
@@ -34154,6 +34190,9 @@ function serializeSession(row) {
     buyerDescription: row.buyerDescription ?? null,
     locations: row.locations ?? null,
     seedRawCandidates: row.seedRawCandidates ?? null,
+    seedProviders: row.seedProviders ?? null,
+    seedProviderCounts: row.seedProviderCounts ?? null,
+    canonicalCount: row.canonicalCount ?? null,
     dedupClusterMinSims: row.dedupClusterMinSims ?? null,
     dedupBandPairFraction: row.dedupBandPairFraction ?? null,
     dedupPairsTotal: row.dedupPairsTotal ?? null,
@@ -34266,6 +34305,7 @@ async function executeDiscovery(opts) {
     project: opts.project,
     icpDescription: opts.icpDescription,
     buyerDescription: opts.buyerDescription,
+    seedProviders: opts.seedProviders,
     locations: opts.locations ?? []
   });
   const { kept: unbrandedCandidates, droppedBranded } = filterBrandedSeedCandidates({
@@ -34297,11 +34337,14 @@ async function executeDiscovery(opts) {
     seedFromAnswerCount: seedResult.fromAnswerCount ?? null,
     seedFromGroundingCount: seedResult.fromGroundingCount ?? null,
     seedBrandFilteredCount: droppedBranded.length,
+    canonicalCount: canonicals.length,
     // Full seed provenance + dedup calibration diagnostics: the raw
     // candidate list (as the seed dep returned it, pre-filter) makes every
     // live session a replayable fixture; the similarity stats are the data
     // the 0.95-threshold decision was missing. No gate reads any of these.
     seedRawCandidates: seedResult.candidates,
+    seedProviders: opts.seedProviders ?? null,
+    seedProviderCounts: seedResult.providerCounts ?? null,
     dedupClusterMinSims: dedupStats.perClusterMinSimilarity,
     dedupBandPairFraction: dedupStats.bandPairFraction,
     dedupPairsTotal: dedupStats.pairsTotal,

package/dist/{chunk-QZQLZL3C.js → chunk-Y26GYDS6.js}

@@ -25,7 +25,7 @@ import {
   trafficConnectVercelRequestSchema,
   trafficConnectWordpressRequestSchema,
   trafficEventKindSchema
-} from "./chunk-WXAHSQOX.js";
+} from "./chunk-ZT7GVLS3.js";
 
 // src/config.ts
 import fs from "fs";
@@ -5566,6 +5566,7 @@ var discoveryRunInputSchema = z2.object({
     // upper bound; this just clarifies the meaning of each knob.
     icpDescription: z2.string().min(1).optional().describe("Free-text ICP description. If omitted, the project must already have spec.icpDescription stored."),
     buyerDescription: z2.string().min(1).optional().describe("Who evaluates or buys the offering, separate from the ICP. When present, every generated query is anchored on this buyer."),
+    seedProviders: z2.array(z2.enum(["gemini", "openai"])).min(1).optional().describe('Which providers generate seed candidates. Omit for the Gemini-only default; ["gemini","openai"] merges both phrasing distributions before dedup.'),
     dedupThreshold: z2.number().min(0).max(1).optional().describe("Cosine similarity threshold for clustering seed candidates. Defaults to 0.85. Lower values dedupe more aggressively."),
     maxProbes: z2.number().int().positive().max(DISCOVERY_MAX_PROBES_CAP).optional().describe(`Max canonical queries to probe in this session. Default 100, hard cap ${DISCOVERY_MAX_PROBES_CAP}.`),
     probeConcurrency: z2.number().int().min(1).max(DISCOVERY_PROBE_CONCURRENCY_CAP).optional().describe(`How many probes may run in parallel. Default 1 (strictly serial), hard cap ${DISCOVERY_PROBE_CONCURRENCY_CAP}. Probe rows are persisted in canonical order regardless of concurrency, so this only shortens wall-clock time.`)

package/dist/{chunk-WXAHSQOX.js → chunk-ZT7GVLS3.js}

@@ -2241,6 +2241,14 @@ var discoverySessionDtoSchema = z17.object({
   /** Dedup calibration: fraction of all pairs in the ambiguous 0.90-0.97 band. */
   dedupBandPairFraction: z17.number().nullable().optional(),
   dedupPairsTotal: z17.number().int().nullable().optional(),
+  /** Seed provider set the session ran with (canonical order); null = legacy /
+   *  Gemini-only default. */
+  seedProviders: z17.array(z17.string()).nullable().optional(),
+  /** Raw candidate count contributed per seed provider. */
+  seedProviderCounts: z17.record(z17.string(), z17.number().int()).nullable().optional(),
+  /** True post-dedup canonical count BEFORE the probe-budget slice (seedCount
+   *  is post-truncation). Null on legacy sessions. */
+  canonicalCount: z17.number().int().nullable().optional(),
   dedupThreshold: z17.number().nullable().optional(),
   probeCount: z17.number().int().nullable().optional(),
   citedCount: z17.number().int().nullable().default(null),
@@ -2317,6 +2325,8 @@ function containsWholeToken(normalized, token) {
   }
   return false;
 }
+var DISCOVERY_SEED_PROVIDERS = ["gemini", "openai"];
+var discoverySeedProviderSchema = z17.enum(DISCOVERY_SEED_PROVIDERS);
 var discoveryRunRequestSchema = z17.object({
   icpDescription: z17.string().min(1).optional(),
   /**
@@ -2326,6 +2336,14 @@ var discoveryRunRequestSchema = z17.object({
    * generic provider comparisons.
    */
   buyerDescription: z17.string().min(1).optional(),
+  /**
+   * Which providers generate seed candidates. Omitted = Gemini-only (the
+   * historical behaviour). Canonicalized (deduped + sorted) so the value is
+   * stable for session-identity comparison. Part of the consolidation
+   * IDENTITY: a different provider set produces a different phrasing
+   * distribution, so it must never reuse another set's session.
+   */
+  seedProviders: z17.array(discoverySeedProviderSchema).min(1).transform((arr) => [...new Set(arr)].sort()).optional(),
   dedupThreshold: z17.number().min(0).max(1).optional(),
   maxProbes: z17.number().int().positive().max(DISCOVERY_MAX_PROBES_CAP).optional(),
   /**