feat(extractors): expand catalog, smoke coverage, and sourcing docs

Adds Arc.dev, BC T-Net, Eluta, iCIMS tenants, QAJobsBoard, and SmartRecruiters
manifests with registry/settings/UI wiring; registers full extractor list in
smoke-extractors and documents supplementary board access paths. Aligns Careerjet
v4 with the url query parameter and fixes strict typing in QAJobsBoard.

Co-authored-by: Cursor <cursoragent@cursor.com>

scripts/smoke-extractors.ts

@@ -1,8 +1,14 @@
 /**
- * Tiny smoke-test for new extractors: imports each manifest, runs it with a
- * minimal context, and prints the count of mapped jobs + a few samples.
+ * Smoke-test helper for extractor manifests: imports each manifest, runs it with a
+ * minimal context, and prints mapped job counts + a sample row.
  *
- * Run from repo root:  npx tsx scripts/smoke-extractors.ts [comma,separated,ids]
+ * Run from repo root:
+ *   npx tsx scripts/smoke-extractors.ts
+ *   npx tsx scripts/smoke-extractors.ts arcdev,icims
+ *   npx tsx scripts/smoke-extractors.ts indeed    # alias → `jobspy` (same manifest)
+ *
+ * Keep `ALL_TARGETS` aligned with every shipped manifest under each
+ * `extractors/<name>/` package (`manifest.ts` or `src/manifest.ts`).
  *
  * Loads repo-root `.env` so keyed extractors match orchestrator behavior (plain
  * `tsx` does not read `.env` automatically).
@@ -14,7 +20,7 @@ import { config as loadEnv } from "dotenv";
 import type {
   ExtractorManifest,
   ExtractorRuntimeContext,
-} from "../shared/src/types/extractors";
+} from "job-ops-shared/types/extractors";
 
 const repoRoot = path.resolve(
   path.dirname(fileURLToPath(import.meta.url)),
@@ -22,54 +28,56 @@ const repoRoot = path.resolve(
 );
 loadEnv({ path: path.join(repoRoot, ".env") });
 
+/** Left column width for log alignment (longest pipeline source id today). */
+const ID_COL = 15;
+
+/** JobSpy serves Indeed / LinkedIn / Glassdoor; CLI filter accepts those ids as aliases. */
+const JOBSPY_SITE_IDS = ["indeed", "linkedin", "glassdoor"] as const;
+
+function expandSmokeFilter(ids: Set<string>): Set<string> {
+  const next = new Set(ids);
+  for (const site of JOBSPY_SITE_IDS) {
+    if (next.has(site)) {
+      next.add("jobspy");
+      break;
+    }
+  }
+  return next;
+}
+
 interface Target {
   id: string;
   importPath: string;
   needs?: string[]; // env vars required to run; skipped if missing
   settings?: Record<string, string>;
+  /** When set, replaces the default smoke search terms (use [] for sources that filter client-side). */
+  searchTerms?: string[];
+  /** Geography passed as `selectedCountry` (must match what each extractor expects). */
+  selectedCountry?: string;
 }
 
 const ALL_TARGETS: Target[] = [
   {
-    id: "jobicy",
-    importPath: "../extractors/jobicy/manifest",
-    settings: { jobicyMaxJobsPerTerm: "10" },
-  },
-  {
-    id: "themuse",
-    importPath: "../extractors/themuse/manifest",
-    settings: { themuseMaxJobsPerTerm: "10" },
-  },
-  {
-    id: "usajobs",
-    importPath: "../extractors/usajobs/manifest",
-    needs: ["USAJOBS_API_KEY", "USAJOBS_USER_AGENT"],
-    settings: { usajobsMaxJobsPerTerm: "10" },
-  },
-  {
-    id: "jooble",
-    importPath: "../extractors/jooble/manifest",
-    needs: ["JOOBLE_API_KEY"],
-    settings: { joobleMaxJobsPerTerm: "10" },
-  },
-  {
-    id: "careerjet",
-    importPath: "../extractors/careerjet/manifest",
-    needs: ["CAREERJET_AFFID", "CAREERJET_REFERER", "CAREERJET_USER_IP"],
-    settings: { careerjetMaxJobsPerTerm: "10" },
-  },
-  {
-    id: "reed",
-    importPath: "../extractors/reed/manifest",
-    needs: ["REED_API_KEY"],
-    settings: { reedMaxJobsPerTerm: "10" },
-  },
-  {
-    id: "lever",
-    importPath: "../extractors/lever/manifest",
+    id: "adzuna",
+    importPath: "../extractors/adzuna/manifest",
+    needs: ["ADZUNA_APP_ID", "ADZUNA_APP_KEY"],
+    selectedCountry: "United States",
     settings: {
-      // Known active public Lever board used purely as a connectivity check.
-      leverCompanies: JSON.stringify(["palantir", "netflix"]),
+      adzunaMaxJobsPerTerm: "10",
+      searchCities: "United States",
+    },
+  },
+  {
+    id: "arbeitnow",
+    importPath: "../extractors/arbeitnow/manifest",
+    settings: { arbeitnowMaxJobsPerTerm: "10" },
+  },
+  {
+    id: "arcdev",
+    importPath: "../extractors/arcdev/manifest",
+    settings: {
+      arcRemoteJobsPaths: JSON.stringify(["/remote-jobs/playwright"]),
+      arcMaxJobsPerPath: "20",
     },
   },
   {
@@ -79,6 +87,40 @@ const ALL_TARGETS: Target[] = [
       ashbyCompanies: JSON.stringify(["ramp", "linear"]),
     },
   },
+  {
+    id: "bctenet",
+    importPath: "../extractors/bctenet/manifest",
+    selectedCountry: "Canada",
+    settings: {
+      bctenetMaxJobsPerTerm: "25",
+    },
+  },
+  {
+    id: "careerjet",
+    importPath: "../extractors/careerjet/manifest",
+    needs: ["CAREERJET_AFFID", "CAREERJET_REFERER", "CAREERJET_USER_IP"],
+    settings: { careerjetMaxJobsPerTerm: "10", searchCities: "United States" },
+  },
+  {
+    id: "eluta",
+    importPath: "../extractors/eluta/manifest",
+    selectedCountry: "Canada",
+    settings: {
+      elutaRssLocations: JSON.stringify(["Toronto, ON"]),
+      elutaMaxJobsPerTerm: "15",
+    },
+  },
+  {
+    id: "fourdayweek",
+    importPath: "../extractors/fourdayweek/manifest",
+    settings: { fourdayweekMaxJobsPerTerm: "10" },
+  },
+  {
+    id: "gradcracker",
+    importPath: "../extractors/gradcracker/manifest",
+    selectedCountry: "United Kingdom",
+    settings: { gradcrackerMaxJobsPerTerm: "10" },
+  },
   {
     id: "greenhouse",
     importPath: "../extractors/greenhouse/manifest",
@@ -87,14 +129,71 @@ const ALL_TARGETS: Target[] = [
     },
   },
   {
-    id: "workday",
-    importPath: "../extractors/workday/manifest",
+    id: "himalayas",
+    importPath: "../extractors/himalayas/manifest",
+    settings: { himalayasMaxJobsPerTerm: "10" },
+  },
+  {
+    id: "hiringcafe",
+    importPath: "../extractors/hiringcafe/manifest",
+    selectedCountry: "United Kingdom",
     settings: {
-      workdayTenants: JSON.stringify([
-        "https://nvidia.wd5.myworkdayjobs.com/en-US/NVIDIAExternalCareerSite",
-      ]),
+      searchCities: "UK",
+      jobspyResultsWanted: "10",
+      workplaceTypes: JSON.stringify(["remote", "hybrid", "onsite"]),
     },
   },
+  {
+    id: "icims",
+    importPath: "../extractors/icims/manifest",
+    searchTerms: [],
+    settings: {
+      icimsTenants: JSON.stringify(["careers-appliedsystems.icims.com"]),
+      icimsMaxJobsPerTenant: "15",
+      icimsMaxPagesPerSearch: "2",
+    },
+  },
+  {
+    id: "jobicy",
+    importPath: "../extractors/jobicy/manifest",
+    settings: { jobicyMaxJobsPerTerm: "10" },
+  },
+  {
+    id: "jobspy",
+    importPath: "../extractors/jobspy/manifest",
+    selectedCountry: "United Kingdom",
+    settings: {
+      searchCities: "UK",
+      jobspyCountryIndeed: "UK",
+      jobspyResultsWanted: "5",
+      workplaceTypes: JSON.stringify(["remote", "hybrid", "onsite"]),
+    },
+  },
+  {
+    id: "jooble",
+    importPath: "../extractors/jooble/manifest",
+    needs: ["JOOBLE_API_KEY"],
+    settings: { joobleMaxJobsPerTerm: "10", searchCities: "United States" },
+  },
+  {
+    id: "lever",
+    importPath: "../extractors/lever/manifest",
+    settings: {
+      leverCompanies: JSON.stringify(["palantir", "netflix"]),
+    },
+  },
+  {
+    id: "qajobsboard",
+    importPath: "../extractors/qajobsboard/manifest",
+    settings: { qajobsboardMaxJobsPerTerm: "25" },
+  },
+  {
+    id: "reed",
+    importPath: "../extractors/reed/manifest",
+    needs: ["REED_API_KEY"],
+    selectedCountry: "United Kingdom",
+    settings: { reedMaxJobsPerTerm: "10" },
+  },
   {
     id: "remoteok",
     importPath: "../extractors/remoteok/manifest",
@@ -106,14 +205,40 @@ const ALL_TARGETS: Target[] = [
     settings: { remotiveMaxJobsPerTerm: "10" },
   },
   {
-    id: "arbeitnow",
-    importPath: "../extractors/arbeitnow/manifest",
-    settings: { arbeitnowMaxJobsPerTerm: "10" },
+    id: "smartrecruiters",
+    importPath: "../extractors/smartrecruiters/manifest",
+    settings: {
+      smartrecruitersCompanies: JSON.stringify(["smartrecruiters"]),
+      smartrecruitersMaxJobsPerCompany: "5",
+    },
   },
   {
-    id: "himalayas",
-    importPath: "../extractors/himalayas/manifest",
-    settings: { himalayasMaxJobsPerTerm: "10" },
+    id: "startupjobs",
+    importPath: "../extractors/startupjobs/src/manifest",
+    selectedCountry: "United Kingdom",
+    settings: {
+      searchCities: "UK",
+      startupjobsMaxJobsPerTerm: "10",
+      workplaceTypes: JSON.stringify(["remote", "hybrid", "onsite"]),
+    },
+  },
+  {
+    id: "themuse",
+    importPath: "../extractors/themuse/manifest",
+    settings: { themuseMaxJobsPerTerm: "10" },
+  },
+  {
+    id: "ukvisajobs",
+    importPath: "../extractors/ukvisajobs/manifest",
+    needs: ["UKVISAJOBS_EMAIL", "UKVISAJOBS_PASSWORD"],
+    selectedCountry: "United Kingdom",
+    settings: { ukvisajobsMaxJobs: "10" },
+  },
+  {
+    id: "usajobs",
+    importPath: "../extractors/usajobs/manifest",
+    needs: ["USAJOBS_API_KEY", "USAJOBS_USER_AGENT"],
+    settings: { usajobsMaxJobsPerTerm: "10" },
   },
   {
     id: "weworkremotely",
@@ -121,22 +246,29 @@ const ALL_TARGETS: Target[] = [
     settings: { weworkremotelyMaxJobsPerTerm: "10" },
   },
   {
-    id: "fourdayweek",
-    importPath: "../extractors/fourdayweek/manifest",
-    settings: { fourdayweekMaxJobsPerTerm: "10" },
+    id: "workday",
+    importPath: "../extractors/workday/manifest",
+    settings: {
+      workdayTenants: JSON.stringify([
+        "https://nvidia.wd5.myworkdayjobs.com/en-US/NVIDIAExternalCareerSite",
+      ]),
+    },
   },
 ];
 
-function buildContext(
-  source: string,
-  settings: Record<string, string>,
-): ExtractorRuntimeContext {
+function buildContext(args: {
+  source: string;
+  settings: Record<string, string>;
+  searchTerms?: string[];
+  selectedCountry?: string;
+}): ExtractorRuntimeContext {
   return {
-    source,
-    selectedSources: [source],
-    settings,
-    searchTerms: ["software engineer"],
-    selectedCountry: "United States",
+    source: args.source,
+    selectedSources: [args.source],
+    settings: args.settings,
+    searchTerms:
+      args.searchTerms !== undefined ? args.searchTerms : ["software engineer"],
+    selectedCountry: args.selectedCountry ?? "United States",
     getExistingJobUrls: async () => [],
     shouldCancel: () => false,
     onProgress: () => {},
@@ -151,7 +283,7 @@ async function runOne(target: Target): Promise<void> {
   const missing = (target.needs ?? []).filter((k) => !process.env[k]);
   if (missing.length > 0) {
     console.log(
-      `${pad(target.id, 12)}  SKIP  missing env: ${missing.join(", ")}`,
+      `${pad(target.id, ID_COL)}  SKIP  missing env: ${missing.join(", ")}`,
     );
     return;
   }
@@ -161,20 +293,25 @@ async function runOne(target: Target): Promise<void> {
     mod = await import(target.importPath);
   } catch (err) {
     console.log(
-      `${pad(target.id, 12)}  FAIL  import error: ${(err as Error).message}`,
+      `${pad(target.id, ID_COL)}  FAIL  import error: ${(err as Error).message}`,
     );
     return;
   }
 
   const manifest = mod.manifest ?? mod.default;
   if (!manifest) {
-    console.log(`${pad(target.id, 12)}  FAIL  manifest export missing`);
+    console.log(`${pad(target.id, ID_COL)}  FAIL  manifest export missing`);
     return;
   }
 
   const started = Date.now();
   try {
-    const ctx = buildContext(target.id, target.settings ?? {});
+    const ctx = buildContext({
+      source: target.id,
+      settings: target.settings ?? {},
+      searchTerms: target.searchTerms,
+      selectedCountry: target.selectedCountry,
+    });
     const result = await manifest.run(ctx);
     const ms = Date.now() - started;
     const status = result.success ? "OK  " : "ERR ";
@@ -183,12 +320,12 @@ async function runOne(target: Target): Promise<void> {
       ? ` | first: "${sample.title}" @ ${sample.employer}`
       : "";
     console.log(
-      `${pad(target.id, 12)}  ${status} jobs=${result.jobs.length}  ${ms}ms${result.error ? ` | error: ${result.error}` : ""}${sampleStr}`,
+      `${pad(target.id, ID_COL)}  ${status} jobs=${result.jobs.length}  ${ms}ms${result.error ? ` | error: ${result.error}` : ""}${sampleStr}`,
     );
   } catch (err) {
     const ms = Date.now() - started;
     console.log(
-      `${pad(target.id, 12)}  CRASH ${ms}ms  ${(err as Error).message}`,
+      `${pad(target.id, ID_COL)}  CRASH ${ms}ms  ${(err as Error).message}`,
     );
   }
 }
@@ -196,11 +333,13 @@ async function runOne(target: Target): Promise<void> {
 async function main() {
   const requested = (process.argv[2] ?? "").trim();
   const filter = requested
-    ? new Set(
-        requested
-          .split(",")
-          .map((s) => s.trim())
-          .filter(Boolean),
+    ? expandSmokeFilter(
+        new Set(
+          requested
+            .split(",")
+            .map((s) => s.trim())
+            .filter(Boolean),
+        ),
       )
     : null;
   const targets = filter