openings 0.1.62 → 0.1.64

package/data/excluded.json

@@ -0,0 +1,63 @@
+{
+  "version": 1,
+  "comment": "Sources that must never enter the catalog, and why. Verification consults this before admitting a board, so a decision taken once is not quietly undone by the next discovery run. Removing a row from data/companies.json does not survive: the lead stays in the registry and the next verification re-admits it.",
+  "entries": [
+    {
+      "ats": "oraclecloud",
+      "token": "ejwl.fa.us2.oraclecloud.com/CX",
+      "name": "Marriott International",
+      "reason": "terms",
+      "detail": "The terms page refuses our requests, and being unable to read a contract is not permission to act under it.",
+      "excludedAt": "2026-09-17"
+    },
+    {
+      "ats": "oraclecloud",
+      "token": "fa-ewjt-saasfaprod1.fa.ocs.oraclecloud.com/CX_2",
+      "name": "EXL",
+      "reason": "terms",
+      "detail": "Terms prohibit automated data gathering.",
+      "excludedAt": "2026-09-17"
+    },
+
+    {
+      "ats": "workable",
+      "token": "eramtalent-1",
+      "name": "Eram Talent",
+      "reason": "staffing",
+      "detail": "A recruiting firm's board lists other companies' vacancies, which is the opposite of the employer's own posting we promise.",
+      "excludedAt": "2026-10-07"
+    },
+    { "ats": "workable", "token": "flextal-staffing-llc", "name": "FlexTal Staffing LLC", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "marathon-talent", "name": "Marathon Talent", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "next-job-abroad", "name": "Next Job Abroad", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "nobox-hr-outsourcing-solutions", "name": "Nobox Aviation Recruitment", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "opendatajobs", "name": "OpenDataJobs", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "osa", "name": "Outsourcing Advantage", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "pairedrecruiting", "name": "Paired", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "pearltalent", "name": "Pearl Talent", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "remote-recruitment", "name": "Remote Recruitment", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "remote-talent-cloud", "name": "Remote Talent Cloud", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "remote-talent-latam", "name": "Remote Talent LATAM", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "staffing-for-doctors", "name": "Staffing for Doctors", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "talent-engine", "name": "Talent Engine", "reason": "staffing", "excludedAt": "2026-10-07" },
+    { "ats": "workable", "token": "uptalent-dot-i-o", "name": "Uptalent.io", "reason": "staffing", "excludedAt": "2026-10-07" },
+
+    {
+      "ats": "workable",
+      "token": "entaingroup",
+      "name": "Entain",
+      "reason": "duplicate",
+      "detail": "Already crawled on SmartRecruiters as `entain`. The same vacancy reached from two boards is two rows in search, which is the one thing the count of unique posts must not include.",
+      "excludedAt": "2026-10-07"
+    },
+    { "ats": "workable", "token": "rentokil-initial", "name": "Rentokil Initial", "reason": "duplicate", "detail": "Already crawled on SmartRecruiters as `rentokilinitial1`.", "excludedAt": "2026-10-07" },
+    {
+      "ats": "workday",
+      "token": "accenture.wd103.myworkdayjobs.com/accenture/AccentureCareers",
+      "name": "Accenture",
+      "reason": "duplicate",
+      "detail": "Accenture's India roles already arrive through the `accenture-india` portal, which is the single largest source in the catalog at 24,400 live roles. The global Workday board overlaps it.",
+      "excludedAt": "2026-10-07"
+    }
+  ]
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "openings",
-  "version": "0.1.62",
+  "version": "0.1.64",
   "description": "A free, candidate-safe job-search substrate for AI agents",
   "license": "MIT",
   "repository": { "type": "git", "url": "git+https://github.com/abhay-avagama/hiring-agent.git" },
@@ -13,6 +13,7 @@
   "files": [
     "src/",
     "data/companies.json",
+    "data/excluded.json",
     "skills/openings/",
     ".codex-plugin/plugin.json",
     ".mcp.json",

package/src/crawl-reporting.ts

@@ -27,8 +27,12 @@ export function resolveAggregatorUrl(value: string | undefined): string | undefi
 }
 
 /** Posts each successfully crawled partition to the aggregator. Only job data is sent, never resume content. */
-export function createCrawlReporter(options: { url: string; fetcher?: Fetch; timeoutMs?: number; attempts?: number; sleep?: (ms: number) => Promise<void> }) {
+export function createCrawlReporter(options: { url: string; fetcher?: Fetch; timeoutMs?: number; attempts?: number; secret?: string; sleep?: (ms: number) => Promise<void> }) {
   const fetcher = options.fetcher ?? globalThis.fetch;
+  // Without this the only thing standing between a stranger and an employer's listings is the slug, ats and
+  // token triple — all three of which are published in data/companies.json inside the npm tarball. An empty
+  // report for a real slug empties that employer from the site and answers 200.
+  const secret = options.secret ?? process.env.OPENINGS_INGEST_SECRET;
   const target = endpoint(options.url, "v1/crawls");
   const attempts = Math.max(1, options.attempts ?? 3);
   const sleep = options.sleep ?? ((ms: number) => new Promise<void>((resolve) => setTimeout(resolve, ms)));
@@ -42,7 +46,8 @@ export function createCrawlReporter(options: { url: string; fetcher?: Fetch; tim
       try {
         const response = await fetcher(target, {
           method: "POST",
-          headers: { "content-type": "application/json", "content-encoding": "gzip" },
+          headers: { "content-type": "application/json", "content-encoding": "gzip",
+                     ...(secret ? { authorization: `Bearer ${secret}` } : {}) },
           body,
           // Big employers send tens of megabytes and the aggregator ingests one report at a time; give them room.
           signal: AbortSignal.timeout(options.timeoutMs ?? 120_000),

package/src/version.ts

@@ -1 +1 @@
-export const VERSION = "0.1.62";
+export const VERSION = "0.1.64";