@hasna/experts 0.0.7 → 0.0.8

package/dist/index.js

@@ -13,6 +13,7 @@ var __export = (target, all) => {
       set: __exportSetter.bind(all, name)
     });
 };
+var __require = import.meta.require;
 // src/db.ts
 import { Database } from "bun:sqlite";
 import { homedir } from "os";
@@ -61,6 +62,36 @@ function authorityScore(e, inputs = {}, weights = DEFAULT_WEIGHTS) {
   const raw = weights.rating * rating + weights.reviews * reviews + weights.followers * followers + weights.featured * featured + weights.verified * verified + weights.recency * recency;
   return Math.round(raw * 1000) / 10;
 }
+function pricePerHour(price, priceUnit) {
+  if (!price || price <= 0)
+    return /free/i.test(priceUnit) ? 0 : null;
+  const u = (priceUnit || "").toLowerCase();
+  const minMatch = u.match(/(\d+)\s*min/);
+  if (minMatch)
+    return Math.round(price * 60 / Number(minMatch[1]));
+  if (/per\s*min|\/\s*min|minute/.test(u))
+    return price * 60;
+  if (/hour|\/\s*hr|per\s*hr/.test(u))
+    return price;
+  if (/free/.test(u))
+    return 0;
+  return null;
+}
+var DEFAULT_BLEND = { semantic: 0.8, authority: 0.2 };
+function blendScore(semantic, authority, w = DEFAULT_BLEND) {
+  const a = Math.max(0, Math.min(1, (authority || 0) / 100));
+  const s = Math.max(0, Math.min(1, semantic));
+  return w.semantic * s + w.authority * a;
+}
+function explainMatch(query, e) {
+  const q = ` ${(query || "").toLowerCase()} `;
+  const hit = (label) => {
+    const l = label.toLowerCase();
+    return q.includes(` ${l} `) || q.includes(`${l},`) || q.includes(`${l}.`) || new RegExp(`\\b${l.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}\\b`).test(q);
+  };
+  const matched = [...e.topics, ...e.tags].filter(hit);
+  return [...new Set(matched)].slice(0, 6);
+}
 
 // src/embed.ts
 var STOPWORDS = new Set([
@@ -151,11 +182,40 @@ class OpenAIEmbedder {
     return data.data.map((d) => d.embedding);
   }
 }
-function getEmbedder() {
-  if (process.env.EXPERTS_EMBEDDER === "openai" && process.env.OPENAI_API_KEY) {
+
+class TransformersEmbedder {
+  id = "minilm-l6-v2";
+  dim = 384;
+  model = process.env.EXPERTS_EMBED_MODEL || "Xenova/all-MiniLM-L6-v2";
+  extractor = null;
+  async ensure() {
+    if (this.extractor)
+      return;
+    const { pipeline } = await import("@huggingface/transformers");
+    this.extractor = await pipeline("feature-extraction", this.model);
+  }
+  async embed(texts) {
+    await this.ensure();
+    const out = [];
+    for (const t of texts) {
+      const r = await this.extractor(t || " ", { pooling: "mean", normalize: true });
+      out.push(Array.from(r.data));
+    }
+    return out;
+  }
+}
+async function getEmbedder() {
+  const choice = process.env.EXPERTS_EMBEDDER;
+  if (choice === "openai" && process.env.OPENAI_API_KEY)
     return new OpenAIEmbedder;
+  if (choice === "hash")
+    return new HashingEmbedder;
+  try {
+    await import("@huggingface/transformers");
+    return new TransformersEmbedder;
+  } catch {
+    return new HashingEmbedder;
   }
-  return new HashingEmbedder;
 }
 function cosine(a, b) {
   let dot = 0;
@@ -251,6 +311,51 @@ function clusterPersons(experts) {
   return out;
 }
 
+// src/crypto.ts
+import { createCipheriv, createDecipheriv, createHmac, scryptSync } from "crypto";
+var PREFIX = "enc1:";
+var cachedKey = null;
+var cachedFrom = null;
+function key2() {
+  const secret = process.env.OPEN_EXPERTS_KEY;
+  if (!secret)
+    return null;
+  if (cachedKey && cachedFrom === secret)
+    return cachedKey;
+  cachedKey = scryptSync(secret, "open-experts/contacts/v1", 32);
+  cachedFrom = secret;
+  return cachedKey;
+}
+function encryptionEnabled() {
+  return !!process.env.OPEN_EXPERTS_KEY;
+}
+function maybeEncrypt(plaintext) {
+  const k = key2();
+  if (!k || plaintext == null)
+    return plaintext;
+  if (plaintext.startsWith(PREFIX))
+    return plaintext;
+  const iv = createHmac("sha256", k).update(plaintext).digest().subarray(0, 12);
+  const cipher = createCipheriv("aes-256-gcm", k, iv);
+  const enc = Buffer.concat([cipher.update(plaintext, "utf8"), cipher.final()]);
+  const tag = cipher.getAuthTag();
+  return PREFIX + Buffer.concat([iv, tag, enc]).toString("base64");
+}
+function maybeDecrypt(stored) {
+  if (stored == null || !stored.startsWith(PREFIX))
+    return stored;
+  const k = key2();
+  if (!k)
+    return stored;
+  const raw = Buffer.from(stored.slice(PREFIX.length), "base64");
+  const iv = raw.subarray(0, 12);
+  const tag = raw.subarray(12, 28);
+  const enc = raw.subarray(28);
+  const decipher = createDecipheriv("aes-256-gcm", k, iv);
+  decipher.setAuthTag(tag);
+  return Buffer.concat([decipher.update(enc), decipher.final()]).toString("utf8");
+}
+
 // src/db.ts
 function defaultDbPath() {
   return process.env.OPEN_EXPERTS_DB || join(homedir(), ".hasna", "experts", "experts.db");
@@ -362,10 +467,11 @@ class ExpertsDB {
       CREATE INDEX IF NOT EXISTS idx_contacts_expert ON contacts(source, source_id);
       CREATE INDEX IF NOT EXISTS idx_contacts_status ON contacts(status);
 
-      -- Semantic search: one embedding vector per expert.
+      -- Semantic search: one embedding vector per expert (text_hash enables
+      -- incremental re-embedding \u2014 skip unchanged experts).
       CREATE TABLE IF NOT EXISTS vectors (
         source TEXT NOT NULL, source_id TEXT NOT NULL,
-        embedder TEXT NOT NULL, dim INTEGER, vec BLOB,
+        embedder TEXT NOT NULL, dim INTEGER, vec BLOB, text_hash TEXT,
         PRIMARY KEY (source, source_id)
       );
 
@@ -387,6 +493,7 @@ class ExpertsDB {
     `);
     this.addColumnIfMissing("experts", "avatar_local", "TEXT");
     this.addColumnIfMissing("experts", "authority", "REAL DEFAULT 0");
+    this.addColumnIfMissing("vectors", "text_hash", "TEXT");
   }
   addColumnIfMissing(table, column, type) {
     const cols = this.db.query(`PRAGMA table_info(${table})`).all();
@@ -473,6 +580,7 @@ class ExpertsDB {
       extra: JSON.parse(r.extra || "{}"),
       avatarLocal: r.avatar_local || undefined,
       authority: r.authority ?? 0,
+      pricePerHour: pricePerHour(r.price ?? 0, r.price_unit ?? ""),
       crawledAt: r.crawled_at
     };
   }
@@ -596,11 +704,11 @@ class ExpertsDB {
     sql += " ORDER BY name";
     return this.db.query(sql).all(...params);
   }
-  setMeta(key2, value) {
-    this.db.query("INSERT INTO meta (key, value) VALUES (?, ?) ON CONFLICT(key) DO UPDATE SET value=excluded.value").run(key2, value);
+  setMeta(key3, value) {
+    this.db.query("INSERT INTO meta (key, value) VALUES (?, ?) ON CONFLICT(key) DO UPDATE SET value=excluded.value").run(key3, value);
   }
-  getMeta(key2) {
-    const row = this.db.query("SELECT value FROM meta WHERE key = ?").get(key2);
+  getMeta(key3) {
+    const row = this.db.query("SELECT value FROM meta WHERE key = ?").get(key3);
     return row ? row.value : null;
   }
   stats(source) {
@@ -624,12 +732,12 @@ class ExpertsDB {
       const nodeIds = new Map;
       const insertNode = this.db.query("INSERT INTO kg_nodes (type, key, label) VALUES (?, ?, ?) ON CONFLICT(type, key) DO UPDATE SET label=excluded.label RETURNING id");
       const insertEdge = this.db.query("INSERT OR REPLACE INTO kg_edges (src, dst, rel, weight) VALUES (?, ?, ?, ?)");
-      const node = (type, key2, label) => {
-        const ck = `${type}\x00${key2.toLowerCase()}`;
+      const node = (type, key3, label) => {
+        const ck = `${type}\x00${key3.toLowerCase()}`;
         const cached = nodeIds.get(ck);
         if (cached != null)
           return cached;
-        const id = insertNode.get(type, key2.toLowerCase(), label).id;
+        const id = insertNode.get(type, key3.toLowerCase(), label).id;
         nodeIds.set(ck, id);
         return id;
       };
@@ -639,7 +747,7 @@ class ExpertsDB {
         for (const topic of e.topics) {
           insertEdge.run(eId, node("topic", topic, topic), "IN_TOPIC", 1);
         }
-        const tweetText = this.recentTweets(e.source, e.sourceId, 30).map((t) => t.text).join(". ");
+        const tweetText = this.recentTweets(e.source, e.sourceId, 30).filter((t) => !t.isRetweet).map((t) => t.text).join(". ");
         const tags = inferTags(expertText(e) + ". " + tweetText, vocabulary);
         for (const tag of tags) {
           insertEdge.run(eId, node("tag", tag, tag), "HAS_TAG", 1);
@@ -693,11 +801,11 @@ class ExpertsDB {
       lastSeen: r.last_seen || ""
     }));
   }
-  expertFromNodeKey(key2) {
-    const idx = key2.indexOf(":");
+  expertFromNodeKey(key3) {
+    const idx = key3.indexOf(":");
     if (idx < 0)
       return null;
-    return this.get(key2.slice(idx + 1), key2.slice(0, idx));
+    return this.get(key3.slice(idx + 1), key3.slice(0, idx));
   }
   findByNeeds(needs, opts = {}) {
     const cleaned = needs.map((n) => n.trim().toLowerCase()).filter(Boolean);
@@ -824,6 +932,15 @@ class ExpertsDB {
     };
   }
   replaceTweets(source, sourceId, tweets) {
+    const norm = (t) => (t || "").toLowerCase().replace(/^rt @\w+:\s*/, "").replace(/https?:\/\/\S+/g, "").replace(/[^a-z0-9 ]/g, "").replace(/\s+/g, " ").trim();
+    const seen = new Set;
+    const deduped = tweets.filter((t) => {
+      const k = norm(t.text);
+      if (!k || seen.has(k))
+        return false;
+      seen.add(k);
+      return true;
+    });
     const tx = this.db.transaction((rows) => {
       this.db.query("DELETE FROM tweets WHERE source = ? AND source_id = ?").run(source, sourceId);
       const stmt = this.db.query(`
@@ -836,7 +953,7 @@ class ExpertsDB {
         stmt.run(t.source, t.sourceId, t.tweetId, t.text, t.createdAt, t.retweetCount, t.replyCount, t.likeCount, t.quoteCount, t.impressionCount, t.isRetweet ? 1 : 0, t.isReply ? 1 : 0);
       }
     });
-    tx(tweets);
+    tx(deduped);
   }
   recentTweets(source, sourceId, limit = 10) {
     const rows = this.db.query("SELECT * FROM tweets WHERE source = ? AND source_id = ? ORDER BY created_at DESC LIMIT ?").all(source, sourceId, limit);
@@ -954,17 +1071,24 @@ class ExpertsDB {
     const log = opts.onLog ?? (() => {});
     const experts = this.list({ source: opts.source });
     const batch = opts.batch ?? 64;
-    const stmt = this.db.query("INSERT OR REPLACE INTO vectors (source, source_id, embedder, dim, vec) VALUES (?, ?, ?, ?, ?)");
+    const stmt = this.db.query("INSERT OR REPLACE INTO vectors (source, source_id, embedder, dim, vec, text_hash) VALUES (?, ?, ?, ?, ?, ?)");
+    const existing = new Map(this.db.query("SELECT source, source_id, embedder, text_hash FROM vectors").all().map((r) => [`${r.source}:${r.source_id}`, { embedder: r.embedder, hash: r.text_hash || "" }]));
+    const work = experts.map((e) => ({ e, text: expertEmbedText(e), hash: String(Bun.hash(expertEmbedText(e))) })).filter(({ e, hash }) => {
+      if (opts.force)
+        return true;
+      const prev = existing.get(`${e.source}:${e.sourceId}`);
+      return !prev || prev.embedder !== embedder.id || prev.hash !== hash;
+    });
     let done = 0;
-    for (let i = 0;i < experts.length; i += batch) {
-      const slice = experts.slice(i, i + batch);
-      const vecs = await embedder.embed(slice.map((e) => expertEmbedText(e)));
+    for (let i = 0;i < work.length; i += batch) {
+      const slice = work.slice(i, i + batch);
+      const vecs = await embedder.embed(slice.map((w) => w.text));
       const tx = this.db.transaction(() => {
-        slice.forEach((e, j) => stmt.run(e.source, e.sourceId, embedder.id, embedder.dim, packVector(vecs[j])));
+        slice.forEach((w, j) => stmt.run(w.e.source, w.e.sourceId, embedder.id, embedder.dim, packVector(vecs[j]), w.hash));
       });
       tx();
       done += slice.length;
-      log(`  embedded ${done}/${experts.length}`);
+      log(`  embedded ${done}/${work.length} (${experts.length - work.length} unchanged)`);
     }
     this.setMeta("embedder", embedder.id);
     this.setMeta("embedded_at", new Date().toISOString());
@@ -977,7 +1101,13 @@ class ExpertsDB {
     const where = opts.source ? "WHERE v.source = ?" : "";
     const params = opts.source ? [opts.source] : [];
     const rows = this.db.query(`SELECT e.*, v.vec AS _vec FROM vectors v JOIN experts e ON e.source=v.source AND e.source_id=v.source_id ${where}`).all(...params);
-    const scored = rows.map((r) => ({ expert: this.rowToExpert(r), score: cosine(queryVec, unpackVector(r._vec)) }));
+    const blend = opts.blend !== false;
+    const scored = rows.map((r) => {
+      const expert = this.rowToExpert(r);
+      const semantic = cosine(queryVec, unpackVector(r._vec));
+      const score = blend ? blendScore(semantic, expert.authority ?? 0) : semantic;
+      return { expert, score, semantic };
+    });
     scored.sort((a, b) => b.score - a.score);
     return scored.slice(0, opts.limit ?? 25);
   }
@@ -1067,7 +1197,7 @@ class ExpertsDB {
       $source: c.source,
       $source_id: c.sourceId,
       $type: c.type,
-      $value: c.value,
+      $value: maybeEncrypt(c.value),
       $label: c.label,
       $provider: c.provider,
       $confidence: c.confidence,
@@ -1077,7 +1207,7 @@ class ExpertsDB {
     });
   }
   setContactStatus(source, sourceId, type, value, status) {
-    this.db.query("UPDATE contacts SET status = ?, verified_at = ? WHERE source = ? AND source_id = ? AND type = ? AND value = ?").run(status, new Date().toISOString(), source, sourceId, type, value);
+    this.db.query("UPDATE contacts SET status = ?, verified_at = ? WHERE source = ? AND source_id = ? AND type = ? AND value = ?").run(status, new Date().toISOString(), source, sourceId, type, maybeEncrypt(value));
   }
   contacts(source, sourceId) {
     const rows = this.db.query("SELECT * FROM contacts WHERE source = ? AND source_id = ? ORDER BY type, confidence DESC").all(source, sourceId);
@@ -1085,7 +1215,7 @@ class ExpertsDB {
       source: r.source,
       sourceId: r.source_id,
       type: r.type,
-      value: r.value,
+      value: maybeDecrypt(r.value),
       label: r.label || "",
       provider: r.provider || "",
       confidence: r.confidence ?? 0,
@@ -1110,7 +1240,7 @@ class ExpertsDB {
       source: r.source,
       sourceId: r.source_id,
       type: r.type,
-      value: r.value,
+      value: maybeDecrypt(r.value),
       label: r.label || "",
       provider: r.provider || "",
       confidence: r.confidence ?? 0,
@@ -1419,34 +1549,37 @@ async function fetchJson(url, fetchFn, init = {}) {
 }
 
 // src/sources/mentorcruise.ts
+function stripHtml(s) {
+  return (s || "").replace(/<[^>]+>/g, " ").replace(/&[a-z#0-9]+;/gi, " ").replace(/\s+/g, " ").trim();
+}
 function normalizeMentor(m, crawledAt) {
-  const slug = m.slug || slugify(m.name || String(m.id ?? ""));
+  const path = m.get_absolute_url || "";
+  const slug = path.match(/\/mentor\/([^/]+)/)?.[1] || slugify(m.get_full_name || String(m.objectID ?? ""));
   const socials = {};
   if (m.twitter)
     socials.twitter = m.twitter.startsWith("http") ? m.twitter : `https://x.com/${m.twitter}`;
   if (m.linkedin)
     socials.linkedin = m.linkedin;
+  const price = m.all_prices?.length ? Math.min(...m.all_prices) : Math.round(m.avg_price_per_call ?? 0);
   return makeExpert({
     source: "mentorcruise",
-    sourceId: String(m.id ?? slug),
+    sourceId: String(m.objectID ?? slug),
     slug,
-    url: `https://mentorcruise.com/mentor/${slug}/`,
-    fullName: m.name ?? [m.first_name, m.last_name].filter(Boolean).join(" "),
-    firstName: m.first_name ?? "",
-    lastName: m.last_name ?? "",
-    title: m.job_title ?? "",
-    bio: m.bio ?? "",
-    avatar: m.avatar ?? m.photo ?? "",
-    price: m.price ?? 0,
-    priceCurrency: m.currency ?? "USD",
-    priceUnit: m.price ? "per month" : "",
-    rating: m.rating ?? 0,
-    ratingCount: m.reviews_count ?? 0,
-    verified: Boolean(m.verified),
+    url: path ? `https://mentorcruise.com${path}` : `https://mentorcruise.com/mentor/${slug}/`,
+    fullName: m.get_full_name ?? "",
+    title: (m.cleaned_job_title ?? []).join(", "),
+    bio: stripHtml(m.bio_formatted ?? ""),
+    avatar: m.get_profile_picture ?? "",
+    price,
+    priceCurrency: "USD",
+    priceUnit: price ? "per month" : "",
+    rating: m.avg_rating_float_one_decimal ?? 0,
+    ratingCount: m.number_of_reviews ?? 0,
     featured: Boolean(m.is_top_mentor),
-    topics: m.categories ?? [],
-    tags: m.skills ?? [],
+    topics: m.get_industries ?? [],
+    tags: m.get_skills ?? [],
     socials,
+    extra: { company: m.company ?? "", location: m.get_location_display ?? "", avgPricePerCall: m.avg_price_per_call ?? 0 },
     crawledAt
   });
 }
@@ -1456,44 +1589,61 @@ class MentorCruiseSource {
   description = "MentorCruise \u2014 long-term mentorship from vetted mentors";
   website = "https://mentorcruise.com";
   fetchFn;
-  apiBase;
+  appId;
+  apiKey;
+  index;
   pageSize;
   constructor(opts = {}) {
     this.fetchFn = opts.fetchFn ?? fetch;
-    this.apiBase = opts.apiBase ?? process.env.MENTORCRUISE_API_BASE ?? "https://mentorcruise.com/api";
-    this.pageSize = opts.pageSize ?? 50;
+    this.appId = opts.appId ?? process.env.MENTORCRUISE_ALGOLIA_APP_ID ?? "YD3XA4V91L";
+    this.apiKey = opts.apiKey ?? process.env.MENTORCRUISE_ALGOLIA_API_KEY ?? "454b55a2e50bc884225318d99b0dad1a";
+    this.index = opts.index ?? process.env.MENTORCRUISE_ALGOLIA_INDEX ?? "MentorProfile_prod";
+    this.pageSize = opts.pageSize ?? 200;
   }
   async crawl(opts = {}) {
     const log = opts.onLog ?? (() => {});
     const crawledAt = new Date().toISOString();
+    const url = `https://${this.appId}-dsn.algolia.net/1/indexes/${this.index}/query`;
     const experts = [];
     const tags = new Set;
-    let offset = 0;
-    for (;; ) {
-      const data = await fetchJson(`${this.apiBase}/mentors/?limit=${this.pageSize}&offset=${offset}`, this.fetchFn);
-      const items = data?.results ?? data?.data ?? (Array.isArray(data) ? data : []);
-      if (!items.length)
+    let page = 0;
+    let pages = 1;
+    while (page < pages) {
+      let data;
+      try {
+        const res = await this.fetchFn(url, {
+          method: "POST",
+          headers: {
+            "X-Algolia-Application-Id": this.appId,
+            "X-Algolia-API-Key": this.apiKey,
+            "Content-Type": "application/json"
+          },
+          body: JSON.stringify({ params: `hitsPerPage=${this.pageSize}&page=${page}` })
+        });
+        if (!res.ok)
+          break;
+        data = await res.json();
+      } catch {
         break;
-      for (const m of items) {
-        const e = normalizeMentor(m, crawledAt);
+      }
+      pages = data.nbPages ?? 1;
+      for (const hit of data.hits ?? []) {
+        const e = normalizeMentor(hit, crawledAt);
         experts.push(e);
         for (const t of e.tags)
           tags.add(t);
       }
-      offset += items.length;
-      log(`  mentorcruise: ${experts.length}`);
+      log(`  mentorcruise: ${experts.length}/${data.nbHits ?? "?"}`);
+      page++;
       if (opts.max && experts.length >= opts.max)
         break;
-      if (items.length < this.pageSize)
-        break;
     }
     if (experts.length === 0) {
-      log("mentorcruise: no public listing reachable (set MENTORCRUISE_API_BASE or provide a fetchFn).");
+      log("mentorcruise: Algolia returned nothing (set MENTORCRUISE_ALGOLIA_* or inject fetchFn).");
     }
-    const topics = [];
     return {
       experts: opts.max ? experts.slice(0, opts.max) : experts,
-      topics,
+      topics: [],
       tags: [...tags].map((name) => ({ name, topic: "" })),
       total: experts.length
     };
@@ -1714,6 +1864,9 @@ async function crawlSource(db, sourceName, opts = {}) {
     throw new Error(`Unknown source "${sourceName}". Run \`experts sources\` to list options.`);
   }
   const data = await source.crawl(opts);
+  if (data.experts.length === 0 && db.count(source.name) > 0) {
+    opts.onLog?.(`\u26A0 ${source.name} returned 0 experts but ${db.count(source.name)} are stored \u2014 possible API drift; not overwriting.`);
+  }
   const changes = db.recordChanges(source.name, data.experts);
   db.upsertExperts(data.experts);
   if (data.topics.length)
@@ -2348,10 +2501,68 @@ function normalizePhone(v) {
     return "+" + cleaned.slice(1).replace(/\D/g, "");
   return cleaned;
 }
+function looksFakePhone(v) {
+  const d = normalizePhone(v).replace(/\D/g, "");
+  if (/^(\d)\1+$/.test(d))
+    return true;
+  if (d === "1234567890" || d === "0123456789")
+    return true;
+  if (/^(\d{2,3})\1\1+$/.test(d))
+    return true;
+  return false;
+}
 function validatePhone(v) {
   const n = normalizePhone(v);
   const digits = n.replace(/\D/g, "");
-  return digits.length >= 7 && digits.length <= 15 ? "valid" : "invalid";
+  if (digits.length < 7 || digits.length > 15)
+    return "invalid";
+  if (looksFakePhone(v))
+    return "invalid";
+  return "valid";
+}
+function interpretRcptCode(code) {
+  if (code === 250 || code === 251)
+    return "valid";
+  if (code === 550 || code === 551 || code === 553 || code === 554)
+    return "invalid";
+  return "unknown";
+}
+async function smtpProbe(email, mxHost) {
+  const net = await import("net");
+  return new Promise((resolve) => {
+    let step = 0;
+    let settled = false;
+    const done = (r) => {
+      if (settled)
+        return;
+      settled = true;
+      try {
+        sock.end();
+      } catch {}
+      resolve(r);
+    };
+    const sock = net.createConnection({ host: mxHost, port: 25, timeout: 7000 });
+    sock.on("timeout", () => done("unknown"));
+    sock.on("error", () => done("unknown"));
+    sock.on("data", (buf) => {
+      const code = parseInt(buf.toString().slice(0, 3), 10);
+      if (step === 0) {
+        sock.write(`HELO open-experts.local\r
+`);
+        step = 1;
+      } else if (step === 1) {
+        sock.write(`MAIL FROM:<probe@open-experts.local>\r
+`);
+        step = 2;
+      } else if (step === 2) {
+        sock.write(`RCPT TO:<${email}>\r
+`);
+        step = 3;
+      } else if (step === 3) {
+        done(interpretRcptCode(code));
+      }
+    });
+  });
 }
 async function validateEmail(email, resolver = resolveMx) {
   const e = normalizeEmail(email);
@@ -2521,12 +2732,25 @@ async function verifyContacts(db, opts = {}) {
   const delayMs = opts.delayMs ?? 50;
   const targets = db.contactsToVerify({ source: opts.source, limit: opts.max });
   const res = { checked: 0, valid: 0, invalid: 0, unknown: 0 };
+  const resolver = opts.resolver;
+  const prober = opts.prober ?? smtpProbe;
   for (const c of targets) {
     let status;
-    if (c.type === "email")
-      status = await validateEmail(c.value, opts.resolver);
-    else
+    if (c.type === "email") {
+      status = await validateEmail(c.value, resolver);
+      if (opts.smtp && status === "valid") {
+        try {
+          const mx = await (resolver ?? (await import("dns/promises")).resolveMx)(c.value.split("@")[1]);
+          if (mx[0]?.exchange) {
+            const probed = await prober(c.value, mx[0].exchange);
+            if (probed !== "unknown")
+              status = probed;
+          }
+        } catch {}
+      }
+    } else {
       status = validatePhone(c.value);
+    }
     db.setContactStatus(c.source, c.sourceId, c.type, c.value, status);
     res.checked++;
     res[status]++;
@@ -2861,7 +3085,7 @@ function _supportsColor(haveStream, { streamIsTTY, sniffFlags = true } = {}) {
     return 1;
   }
   if ("CI" in env) {
-    if (["GITHUB_ACTIONS", "GITEA_ACTIONS", "CIRCLECI"].some((key2) => (key2 in env))) {
+    if (["GITHUB_ACTIONS", "GITEA_ACTIONS", "CIRCLECI"].some((key3) => (key3 in env))) {
       return 3;
     }
     if (["TRAVIS", "APPVEYOR", "GITLAB_CI", "BUILDKITE", "DRONE"].some((sign) => (sign in env)) || env.CI_NAME === "codeship") {
@@ -3262,16 +3486,22 @@ export {
   tokenize,
   syncContacts,
   socialHandles,
+  smtpProbe,
   slugFromUrl,
   samePerson,
   registerSource,
+  pricePerHour,
   normalizePhone,
   normalizeName,
   normalizeIntroExpert,
   normalizeEmail,
+  maybeEncrypt,
+  maybeDecrypt,
+  looksFakePhone,
   liveTweets,
   listSources,
   linkedinHandle,
+  interpretRcptCode,
   inferTags,
   htmlToText,
   handleFromSocial,
@@ -3280,6 +3510,7 @@ export {
   exports_format as format,
   extractJson,
   extractExaContacts,
+  explainMatch,
   expertToContactRecord,
   expertText,
   expertEmbedText,
@@ -3288,6 +3519,7 @@ export {
   enrichSite,
   enrichLinkedIn,
   enrichExpert,
+  encryptionEnabled,
   downloadAvatar,
   discoverContacts,
   defaultRunner,
@@ -3295,9 +3527,11 @@ export {
   crawlSource,
   cosine,
   clusterPersons,
+  blendScore,
   backfillAvatars,
   avatarBasename,
   authorityScore,
+  TransformersEmbedder,
   OpenAIEmbedder,
   JsonSink,
   IntroSource,
@@ -3306,6 +3540,7 @@ export {
   ExpertsDB,
   ExaWebsetsProvider,
   DEFAULT_WEIGHTS,
+  DEFAULT_BLEND,
   CredentialPool,
   ConnectorsClient,
   CliSink

package/dist/score.d.ts

@@ -22,4 +22,32 @@ export interface ScoreInputs {
  * reach, featured/verified flags, and recency of activity.
  */
 export declare function authorityScore(e: Expert, inputs?: ScoreInputs, weights?: ScoreWeights): number;
+/**
+ * Normalize a price + unit to a comparable USD/hour figure so prices can be
+ * compared across sources (intro=per-15-min, Clarity=per-minute, etc.).
+ * Returns null when the unit isn't an hourly-comparable rate (e.g. per-month).
+ */
+export declare function pricePerHour(price: number, priceUnit: string): number | null;
+export interface BlendWeights {
+    /** Weight on semantic similarity (0..1 cosine). */
+    semantic: number;
+    /** Weight on the expert's authority (0..100, normalized here). */
+    authority: number;
+}
+export declare const DEFAULT_BLEND: BlendWeights;
+/**
+ * Blend a semantic similarity score with an expert's authority so ranking isn't
+ * driven by text match alone — a strong text match from an unrated, no-review
+ * advisor shouldn't outrank a verified top expert with a near-equal match.
+ * Returns 0..1.
+ */
+export declare function blendScore(semantic: number, authority: number, w?: BlendWeights): number;
+/**
+ * Explain why an expert matched a free-text query: which of their tags/topics
+ * appear in the query (case-insensitive whole-token overlap). Pure + testable.
+ */
+export declare function explainMatch(query: string, e: {
+    tags: string[];
+    topics: string[];
+}): string[];
 //# sourceMappingURL=score.d.ts.map

package/dist/score.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"score.d.ts","sourceRoot":"","sources":["../src/score.ts"],"names":[],"mappings":"AAAA;;;GAGG;AACH,OAAO,KAAK,EAAE,MAAM,EAAY,MAAM,SAAS,CAAC;AAEhD,MAAM,WAAW,YAAY;IAC3B,MAAM,EAAE,MAAM,CAAC;IACf,OAAO,EAAE,MAAM,CAAC;IAChB,SAAS,EAAE,MAAM,CAAC;IAClB,QAAQ,EAAE,MAAM,CAAC;IACjB,QAAQ,EAAE,MAAM,CAAC;IACjB,OAAO,EAAE,MAAM,CAAC;CACjB;AAED,eAAO,MAAM,eAAe,EAAE,YAO7B,CAAC;AAMF,MAAM,WAAW,WAAW;IAC1B,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,wDAAwD;IACxD,kBAAkB,CAAC,EAAE,MAAM,CAAC;CAC7B;AAED;;;GAGG;AACH,wBAAgB,cAAc,CAAC,CAAC,EAAE,MAAM,EAAE,MAAM,GAAE,WAAgB,EAAE,OAAO,GAAE,YAA8B,GAAG,MAAM,CAkBnH"}
+{"version":3,"file":"score.d.ts","sourceRoot":"","sources":["../src/score.ts"],"names":[],"mappings":"AAAA;;;GAGG;AACH,OAAO,KAAK,EAAE,MAAM,EAAY,MAAM,SAAS,CAAC;AAEhD,MAAM,WAAW,YAAY;IAC3B,MAAM,EAAE,MAAM,CAAC;IACf,OAAO,EAAE,MAAM,CAAC;IAChB,SAAS,EAAE,MAAM,CAAC;IAClB,QAAQ,EAAE,MAAM,CAAC;IACjB,QAAQ,EAAE,MAAM,CAAC;IACjB,OAAO,EAAE,MAAM,CAAC;CACjB;AAED,eAAO,MAAM,eAAe,EAAE,YAO7B,CAAC;AAMF,MAAM,WAAW,WAAW;IAC1B,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,wDAAwD;IACxD,kBAAkB,CAAC,EAAE,MAAM,CAAC;CAC7B;AAED;;;GAGG;AACH,wBAAgB,cAAc,CAAC,CAAC,EAAE,MAAM,EAAE,MAAM,GAAE,WAAgB,EAAE,OAAO,GAAE,YAA8B,GAAG,MAAM,CAkBnH;AAED;;;;GAIG;AACH,wBAAgB,YAAY,CAAC,KAAK,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAS5E;AAED,MAAM,WAAW,YAAY;IAC3B,mDAAmD;IACnD,QAAQ,EAAE,MAAM,CAAC;IACjB,kEAAkE;IAClE,SAAS,EAAE,MAAM,CAAC;CACnB;AAED,eAAO,MAAM,aAAa,EAAE,YAAgD,CAAC;AAE7E;;;;;GAKG;AACH,wBAAgB,UAAU,CAAC,QAAQ,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,CAAC,GAAE,YAA4B,GAAG,MAAM,CAIvG;AAED;;;GAGG;AACH,wBAAgB,YAAY,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,EAAE;IAAE,IAAI,EAAE,MAAM,EAAE,CAAC;IAAC,MAAM,EAAE,MAAM,EAAE,CAAA;CAAE,GAAG,MAAM,EAAE,CAQ7F"}

package/dist/sdk.js

@@ -13,6 +13,7 @@ var __export = (target, all) => {
       set: __exportSetter.bind(all, name)
     });
 };
+var __require = import.meta.require;
 
 // src/sdk.ts
 class ExpertsClient {