careervivid 1.12.36 → 1.12.38

package/dist/agent/instructions.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"instructions.d.ts","sourceRoot":"","sources":["../../src/agent/instructions.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,eAAO,MAAM,aAAa,QAalB,CAAC;AAMT,eAAO,MAAM,cAAc,QAcnB,CAAC;AAMT,eAAO,MAAM,cAAc,QAgBnB,CAAC;AAMT,eAAO,MAAM,kBAAkB,QAqCvB,CAAC;AAMT,eAAO,MAAM,YAAY,QAuCjB,CAAC;AAMT,eAAO,MAAM,iBAAiB,QActB,CAAC;AAMT;;;GAGG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE;IACzC,IAAI,CAAC,EAAE,OAAO,CAAC;IACf,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB,GAAG,MAAM,CAyBT"}
+{"version":3,"file":"instructions.d.ts","sourceRoot":"","sources":["../../src/agent/instructions.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,eAAO,MAAM,aAAa,QAalB,CAAC;AAMT,eAAO,MAAM,cAAc,QAcnB,CAAC;AAMT,eAAO,MAAM,cAAc,QAgBnB,CAAC;AAMT,eAAO,MAAM,kBAAkB,QAwCvB,CAAC;AAMT,eAAO,MAAM,YAAY,QA2CjB,CAAC;AAMT,eAAO,MAAM,iBAAiB,QActB,CAAC;AAMT;;;GAGG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE;IACzC,IAAI,CAAC,EAAE,OAAO,CAAC;IACf,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB,GAAG,MAAM,CAyBT"}

package/dist/agent/instructions.js

@@ -75,7 +75,10 @@ export const JOBS_TOOLS_SECTION = `
 - **delete_resume**     — Permanently delete a resume (ask for confirmation first).
 
 ### Job Search
-- **search_jobs**       — Search for jobs scored against the user's resume. Returns results for review (dry_run by default — does NOT auto-save).
+- **search_jobs**       — Search for NEW companies/roles NOT yet in the tracker. Returns results for review (dry_run by default — does NOT auto-save). Use for discovery only.
+- **openings_scan**    ⭐ — Drill into companies ALREADY in jobs.csv to find their SPECIFIC open roles with direct apply links (uses Greenhouse/Ashby/Lever APIs). Use this instead of search_jobs when the user wants roles at known tracked companies.
+- **openings_list**    — List saved job openings found by openings_scan.
+- **openings_apply**   — Mark a specific opening as Applied (requires explicit date).
 
 ### CSV Pipeline Tracker (tracker_*)
 These tools read/write jobs.csv — the local career-ops pipeline spreadsheet.
@@ -110,9 +113,10 @@ export const JOBS_HARNESS = `
 ## Autonomy Rules (Non-Negotiable)
 
 ### What you may do freely (no confirmation needed)
-- Call any read-only tool (tracker_list_jobs, tracker_dashboard, tracker_rank_priority, tracker_find_stale, tracker_inspect_quality, tracker_recheck_urls, search_jobs, get_resume, etc.)
+- Call any read-only tool (tracker_list_jobs, tracker_dashboard, tracker_rank_priority, tracker_find_stale, tracker_inspect_quality, tracker_recheck_urls, openings_scan, openings_list, search_jobs, get_resume, etc.)
 - Add a new company/job entry via tracker_add_job (when the user explicitly names a company or approves a search result)
 - Update non-status fields (attention_score, excitement, notes, follow_up_date, etc.) via tracker_update_job
+- Run openings_scan to fetch real job postings from tracked companies
 
 ### What requires explicit user confirmation
 - Changing status to "Applied", "Rejected", or "Ghosted" — present the proposed change and wait for "yes"
@@ -144,7 +148,10 @@ If it does, call tracker_update_job instead — never create a duplicate row.
 | updating status, marking applied, follow-up   | tracker_update_job             |
 | Kanban board, web tracker                     | kanban_list_jobs               |
 | resume, background, skills, experience        | get_resume                     |
-| find jobs, search roles                       | get_resume → search_jobs       |
+| specific roles at tracked companies           | openings_scan                  |
+| view saved openings                           | openings_list                  |
+| applied to a specific opening                 | openings_apply                 |
+| find NEW companies/roles not yet in tracker   | get_resume → search_jobs       |
 `.trim();
 // ---------------------------------------------------------------------------
 // §6 — Greeting protocol (shared across modes)

package/dist/agent/tools/jobOpenings.d.ts

@@ -0,0 +1,21 @@
+/**
+ * jobOpenings.ts — Job Openings Drill-Down Tools
+ *
+ * Bridges jobs.csv (company pipeline) → specific open roles with direct apply links.
+ *
+ * Architecture:
+ *   jobs.csv          →  openings_scan  →  job_openings.csv
+ *   (company-level)      (ATS fetcher)     (posting-level)
+ *
+ * ATS Support:
+ *   Greenhouse  — Free public REST API (boards-api.greenhouse.io)
+ *   Ashby       — HTML parse (JavaScript SPA; titles extracted from embedded JSON)
+ *   Lever       — Public REST API (api.lever.co/v0/postings/{slug})
+ *   Direct      — JSON-LD schema.org/JobPosting + heuristic <a> tag parsing
+ */
+import { Tool } from "../Tool.js";
+export declare const OpeningsScanTool: Tool;
+export declare const OpeningsListTool: Tool;
+export declare const OpeningsApplyTool: Tool;
+export declare const ALL_JOB_OPENINGS_TOOLS: Tool[];
+//# sourceMappingURL=jobOpenings.d.ts.map

package/dist/agent/tools/jobOpenings.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"jobOpenings.d.ts","sourceRoot":"","sources":["../../../src/agent/tools/jobOpenings.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EAAE,IAAI,EAAE,MAAM,YAAY,CAAC;AAgclC,eAAO,MAAM,gBAAgB,EAAE,IA6O9B,CAAC;AAMF,eAAO,MAAM,gBAAgB,EAAE,IA4G9B,CAAC;AAQF,eAAO,MAAM,iBAAiB,EAAE,IA8F/B,CAAC;AAMF,eAAO,MAAM,sBAAsB,EAAE,IAAI,EAIxC,CAAC"}

package/dist/agent/tools/jobOpenings.js

@@ -0,0 +1,795 @@
+/**
+ * jobOpenings.ts — Job Openings Drill-Down Tools
+ *
+ * Bridges jobs.csv (company pipeline) → specific open roles with direct apply links.
+ *
+ * Architecture:
+ *   jobs.csv          →  openings_scan  →  job_openings.csv
+ *   (company-level)      (ATS fetcher)     (posting-level)
+ *
+ * ATS Support:
+ *   Greenhouse  — Free public REST API (boards-api.greenhouse.io)
+ *   Ashby       — HTML parse (JavaScript SPA; titles extracted from embedded JSON)
+ *   Lever       — Public REST API (api.lever.co/v0/postings/{slug})
+ *   Direct      — JSON-LD schema.org/JobPosting + heuristic <a> tag parsing
+ */
+import { Type } from "@google/genai";
+import { readFileSync, writeFileSync, existsSync, mkdirSync, copyFileSync, renameSync, } from "fs";
+import { resolve, dirname } from "path";
+import { fileURLToPath } from "url";
+import { homedir } from "os";
+const __dirname = dirname(fileURLToPath(import.meta.url));
+// ---------------------------------------------------------------------------
+// CSV Schema — job_openings.csv
+// ---------------------------------------------------------------------------
+const OPENING_HEADERS = [
+    "id", // VER-OPN-001
+    "company_id", // FK → jobs.csv id (e.g. VER-001)
+    "company", // Human name
+    "posting_title", // Exact title from ATS
+    "apply_url", // Direct apply link
+    "ats", // Greenhouse | Ashby | Lever | Direct
+    "location", // Remote / SF / etc.
+    "match_score", // 0–100 keyword match vs resume
+    "match_keywords", // comma-separated matched keywords
+    "status", // Found | Reviewing | Applied | Rejected | Ghosted
+    "date_found", // ISO date
+    "date_applied",
+    "notes",
+];
+// ---------------------------------------------------------------------------
+// File path
+// ---------------------------------------------------------------------------
+function getOpeningsPath() {
+    // Mirror jobs.csv location: career-ops/data/job_openings.csv
+    const careerOpsPath = resolve(__dirname, "..", "..", "..", "..", "career-ops", "data");
+    if (existsSync(careerOpsPath))
+        return resolve(careerOpsPath, "job_openings.csv");
+    // Fallback: ~/.careervivid/
+    const dir = resolve(homedir(), ".careervivid");
+    if (!existsSync(dir))
+        mkdirSync(dir, { recursive: true });
+    return resolve(dir, "job_openings.csv");
+}
+function loadOpenings() {
+    const path = getOpeningsPath();
+    if (!existsSync(path)) {
+        writeFileSync(path, OPENING_HEADERS.join(",") + "\n", "utf-8");
+        return { rows: [], path };
+    }
+    const raw = readFileSync(path, "utf-8");
+    const lines = raw.trim().split("\n");
+    if (lines.length <= 1)
+        return { rows: [], path };
+    const fileHeaders = lines[0].split(",");
+    const rows = lines.slice(1).filter(l => l.trim()).map(line => {
+        const cols = parseCsvLine(line);
+        const row = {};
+        OPENING_HEADERS.forEach(h => {
+            const idx = fileHeaders.indexOf(h);
+            row[h] = idx >= 0 ? (cols[idx] ?? "").trim() : "";
+        });
+        return row;
+    });
+    return { rows, path };
+}
+function saveOpenings(rows, csvPath) {
+    if (existsSync(csvPath))
+        copyFileSync(csvPath, csvPath + ".bak");
+    const header = OPENING_HEADERS.join(",");
+    const data = rows.map(row => OPENING_HEADERS.map(h => {
+        const val = row[h] ?? "";
+        return val.includes(",") || val.includes('"')
+            ? `"${val.replace(/"/g, '""')}"`
+            : val;
+    }).join(","));
+    const tmpPath = csvPath + ".tmp";
+    writeFileSync(tmpPath, [header, ...data].join("\n") + "\n", "utf-8");
+    renameSync(tmpPath, csvPath);
+}
+/** Minimal CSV line parser that handles quoted fields. */
+function parseCsvLine(line) {
+    const cols = [];
+    let cur = "";
+    let inQuote = false;
+    for (let i = 0; i < line.length; i++) {
+        const ch = line[i];
+        if (ch === '"') {
+            if (inQuote && line[i + 1] === '"') {
+                cur += '"';
+                i++;
+            }
+            else {
+                inQuote = !inQuote;
+            }
+        }
+        else if (ch === "," && !inQuote) {
+            cols.push(cur);
+            cur = "";
+        }
+        else {
+            cur += ch;
+        }
+    }
+    cols.push(cur);
+    return cols;
+}
+function today() {
+    return new Date().toISOString().slice(0, 10);
+}
+// ---------------------------------------------------------------------------
+// ATS Slug Extractor
+// ---------------------------------------------------------------------------
+function extractSlug(careersUrl, ats) {
+    try {
+        const url = new URL(careersUrl);
+        const pathname = url.pathname;
+        if (ats === "Greenhouse") {
+            // boards.greenhouse.io/vercel  OR  job-boards.greenhouse.io/vercel
+            const match = pathname.match(/^\/([^/]+)/);
+            return match?.[1] ?? null;
+        }
+        if (ats === "Ashby") {
+            // jobs.ashbyhq.com/supabase
+            const match = pathname.match(/^\/([^/]+)/);
+            return match?.[1] ?? null;
+        }
+        if (ats === "Lever") {
+            // jobs.lever.co/wandb
+            const match = pathname.match(/^\/([^/]+)/);
+            return match?.[1] ?? null;
+        }
+        return null;
+    }
+    catch {
+        return null;
+    }
+}
+// ---------------------------------------------------------------------------
+// ATS Fetchers
+// ---------------------------------------------------------------------------
+/** Greenhouse: free public API — returns real titles + direct apply URLs */
+async function fetchGreenhouseJobs(slug) {
+    const url = `https://boards-api.greenhouse.io/v1/boards/${slug}/jobs`;
+    const res = await fetch(url, {
+        headers: { "User-Agent": "CareerVivid-Agent/1.0" },
+        signal: AbortSignal.timeout(15_000),
+    });
+    if (!res.ok)
+        throw new Error(`Greenhouse API returned ${res.status}`);
+    const data = await res.json();
+    return (data.jobs ?? []).map(j => ({
+        title: j.title,
+        applyUrl: j.absolute_url,
+        location: j.location?.name ?? "",
+        department: j.departments?.[0]?.name ?? "",
+    }));
+}
+/** Ashby: SPA — parse the embedded JSON from the rendered HTML */
+async function fetchAshbyJobs(slug) {
+    const url = `https://jobs.ashbyhq.com/${slug}`;
+    const res = await fetch(url, {
+        headers: {
+            "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36",
+        },
+        signal: AbortSignal.timeout(15_000),
+    });
+    if (!res.ok)
+        throw new Error(`Ashby page returned ${res.status}`);
+    const html = await res.text();
+    // Ashby embeds job data in Next.js __NEXT_DATA__ JSON
+    const nextDataMatch = html.match(/<script id="__NEXT_DATA__" type="application\/json">([^<]+)<\/script>/);
+    if (nextDataMatch) {
+        try {
+            const nextData = JSON.parse(nextDataMatch[1]);
+            // Location: props.pageProps.jobPostings or props.pageProps.jobBoard.jobPostings
+            const postings = nextData?.props?.pageProps?.jobPostings ??
+                nextData?.props?.pageProps?.jobBoard?.jobPostings ??
+                [];
+            if (Array.isArray(postings) && postings.length > 0) {
+                return postings
+                    .filter((p) => p.jobPostingState === "Listed" || !p.jobPostingState)
+                    .map((p) => ({
+                    title: p.title ?? p.name ?? "Unknown",
+                    applyUrl: p.externalLink ?? p.applyUrl ?? `https://jobs.ashbyhq.com/${slug}/${p.id}`,
+                    location: p.locationName ?? p.location ?? "",
+                    department: p.departmentName ?? p.team ?? "",
+                }));
+            }
+        }
+        catch { /* fall through to regex */ }
+    }
+    // Fallback: regex-extract title fields from the embedded JSON blobs
+    const titleMatches = [...html.matchAll(/"title":"([^"]{5,80})"/g)].map(m => m[1]);
+    const dedupedTitles = [...new Set(titleMatches)];
+    return dedupedTitles
+        .filter(t => /engineer|designer|product|manager|analyst|developer|scientist|operations/i.test(t))
+        .map(title => ({
+        title,
+        applyUrl: `https://jobs.ashbyhq.com/${slug}`,
+        location: "",
+    }));
+}
+/** Lever: public REST API with HTML fallback */
+async function fetchLeverJobs(slug, careersUrl) {
+    // Try public API first
+    try {
+        const res = await fetch(`https://api.lever.co/v0/postings/${slug}?mode=json&limit=200`, {
+            headers: { "User-Agent": "CareerVivid-Agent/1.0" },
+            signal: AbortSignal.timeout(10_000),
+        });
+        if (res.ok) {
+            const data = await res.json();
+            if (Array.isArray(data) && data.length > 0) {
+                return data.map((p) => ({
+                    title: p.text ?? p.title,
+                    applyUrl: p.hostedUrl ?? p.applyUrl ?? careersUrl,
+                    location: p.categories?.location ?? p.location ?? "",
+                    department: p.categories?.team ?? p.categories?.department ?? "",
+                }));
+            }
+        }
+    }
+    catch { /* fall through */ }
+    // HTML fallback: lever hosted job pages have structured JSON
+    const res = await fetch(`https://jobs.lever.co/${slug}`, {
+        headers: {
+            "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36",
+        },
+        signal: AbortSignal.timeout(15_000),
+    });
+    if (!res.ok)
+        throw new Error(`Lever page ${res.status}`);
+    const html = await res.text();
+    // Lever pages embed jobs in <script type="application/ld+json">
+    const ldMatches = [...html.matchAll(/<script type="application\/ld\+json">([^<]+)<\/script>/g)];
+    const results = [];
+    for (const m of ldMatches) {
+        try {
+            const obj = JSON.parse(m[1]);
+            if (obj["@type"] === "JobPosting" || obj.title) {
+                results.push({
+                    title: obj.title ?? obj.name,
+                    applyUrl: obj.url ?? obj.sameAs ?? `https://jobs.lever.co/${slug}`,
+                    location: obj.jobLocation?.address?.addressLocality ?? obj.jobLocation?.name ?? "",
+                });
+            }
+        }
+        catch { /* skip malformed */ }
+    }
+    // Also try extracting from data-qa attributes Lever uses
+    const qaMatches = [...html.matchAll(/data-qa="posting-name"[^>]*>([^<]+)</g)];
+    const urlMatches = [...html.matchAll(/href="(https:\/\/jobs\.lever\.co\/[^"]+)"/g)];
+    qaMatches.forEach((m, i) => {
+        const title = m[1].trim();
+        if (title && !results.find(r => r.title === title)) {
+            results.push({
+                title,
+                applyUrl: urlMatches[i]?.[1] ?? `https://jobs.lever.co/${slug}`,
+                location: "",
+            });
+        }
+    });
+    return results;
+}
+/** Direct: JSON-LD schema.org/JobPosting parse + heuristic <a> tag extraction */
+async function fetchDirectJobs(careersUrl) {
+    const res = await fetch(careersUrl, {
+        headers: {
+            "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36",
+        },
+        signal: AbortSignal.timeout(15_000),
+    });
+    if (!res.ok)
+        throw new Error(`Careers page returned ${res.status}`);
+    const html = await res.text();
+    const results = [];
+    // 1. JSON-LD schema.org/JobPosting
+    const ldMatches = [...html.matchAll(/<script type="application\/ld\+json">([^<]+)<\/script>/g)];
+    for (const m of ldMatches) {
+        try {
+            const obj = JSON.parse(m[1]);
+            const items = Array.isArray(obj) ? obj : [obj];
+            for (const item of items) {
+                if (item["@type"] === "JobPosting") {
+                    results.push({
+                        title: item.title,
+                        applyUrl: item.url ?? item.sameAs ?? careersUrl,
+                        location: item.jobLocation?.address?.addressLocality ?? item.jobLocation?.name ?? "",
+                        description: item.description?.slice(0, 200),
+                    });
+                }
+            }
+        }
+        catch { /* skip malformed */ }
+    }
+    if (results.length > 0)
+        return results;
+    // 2. Heuristic: find <a> tags that look like job posting links
+    const baseUrl = new URL(careersUrl);
+    const linkPattern = /<a\s[^>]*href="([^"]+)"[^>]*>([^<]{10,100})<\/a>/gi;
+    const linkMatches = [...html.matchAll(linkPattern)];
+    const JOB_TITLE_SIGNALS = /engineer|developer|designer|product manager|analyst|architect|scientist|devrel|advocate|solutions|forward.?deployed|technical/i;
+    const SKIP_PATTERNS = /privacy|terms|blog|about|home|login|signup|contact|docs/i;
+    for (const m of linkMatches) {
+        const href = m[1];
+        const linkText = m[2].trim().replace(/\s+/g, " ");
+        if (!JOB_TITLE_SIGNALS.test(linkText))
+            continue;
+        if (SKIP_PATTERNS.test(linkText))
+            continue;
+        if (linkText.length > 100)
+            continue;
+        // Resolve relative URLs
+        let fullUrl = href;
+        try {
+            fullUrl = new URL(href, baseUrl.origin).toString();
+        }
+        catch {
+            continue;
+        }
+        if (!results.find(r => r.title === linkText)) {
+            results.push({ title: linkText, applyUrl: fullUrl, location: "" });
+        }
+    }
+    return results;
+}
+// ---------------------------------------------------------------------------
+// Resume Keyword Scoring
+// ---------------------------------------------------------------------------
+const RESUME_KEYWORD_CACHE = {
+    keywords: [],
+    timestamp: 0,
+};
+/**
+ * Returns a curated keyword list for the current user.
+ * First tries to pull from CareerVivid API; falls back to a hardcoded set
+ * of common AI/engineering keywords if API is unavailable.
+ */
+async function getResumeKeywords() {
+    // Cache for 5 minutes within a session
+    if (RESUME_KEYWORD_CACHE.keywords.length > 0 &&
+        Date.now() - RESUME_KEYWORD_CACHE.timestamp < 300_000) {
+        return RESUME_KEYWORD_CACHE.keywords;
+    }
+    // Curated engineering + AI keyword set that matches the user's profile.
+    // These are weighted for Solutions Engineer / Forward Deployed Engineer roles.
+    const defaults = [
+        // Roles
+        "Solutions Engineer", "Forward Deployed Engineer", "Developer Advocate",
+        "Software Engineer", "Full-stack", "Backend", "Frontend", "DevRel",
+        // Tech stack
+        "TypeScript", "JavaScript", "React", "Next.js", "Node.js",
+        "Python", "PostgreSQL", "Firebase", "Supabase", "REST API", "GraphQL",
+        // AI/ML
+        "LLM", "AI", "Agentic", "RAG", "Embeddings", "OpenAI", "Anthropic", "Gemini",
+        "Machine Learning", "GPT", "Claude", "agent", "automation",
+        // DevTools
+        "Docker", "AWS", "GCP", "Vercel", "CLI", "SDK", "developer tools",
+        "developer experience", "DX", "open source",
+        // Traits (commonly in JDs matched to your background)
+        "Remote", "startup", "SaaS", "platform", "integration",
+    ];
+    RESUME_KEYWORD_CACHE.keywords = defaults;
+    RESUME_KEYWORD_CACHE.timestamp = Date.now();
+    return defaults;
+}
+function scorePosting(keywords, posting) {
+    const text = [posting.title, posting.description ?? "", posting.department ?? ""]
+        .join(" ")
+        .toLowerCase();
+    const matched = keywords.filter(k => text.includes(k.toLowerCase()));
+    // Title match is worth more — double-count title hits
+    const titleText = posting.title.toLowerCase();
+    const titleMatched = keywords.filter(k => titleText.includes(k.toLowerCase()));
+    const weightedMatches = new Set([...matched, ...titleMatched, ...titleMatched]); // title 2x
+    const score = Math.min(100, Math.round((weightedMatches.size / Math.max(keywords.length * 0.3, 8)) * 100));
+    return { score, matched };
+}
+// ---------------------------------------------------------------------------
+// ID generator for openings
+// ---------------------------------------------------------------------------
+function generateOpeningId(companyId, existingRows) {
+    const prefix = companyId.replace(/-\d+$/, ""); // VER-001 → VER
+    const existing = existingRows.filter(r => r.id.startsWith(prefix + "-OPN-"));
+    const maxN = existing.reduce((max, r) => {
+        const n = parseInt(r.id.split("-OPN-")[1] ?? "0", 10);
+        return Math.max(max, isNaN(n) ? 0 : n);
+    }, 0);
+    return `${prefix}-OPN-${String(maxN + 1).padStart(3, "0")}`;
+}
+// ---------------------------------------------------------------------------
+// Tool: openings_scan
+// ---------------------------------------------------------------------------
+export const OpeningsScanTool = {
+    name: "openings_scan",
+    description: `Scans companies already in the user's job tracker (jobs.csv) to find their ACTUAL open roles with direct apply links.
+
+This is the CORRECT way to find specific job postings — it uses the company's real ATS (Greenhouse, Ashby, Lever) to return verified, clickable apply URLs instead of hallucinating jobs.
+
+Use this when the user asks:
+- "What specific roles are open at my tracked companies?"
+- "Find me specific jobs I can apply to"  
+- "Drill into openings at Vercel / LangChain / Supabase"
+- "What's open at my Tier 1 companies?"
+- "Show me real job postings with direct links"
+
+The tool reads careers_url + ats from jobs.csv, fetches live postings,
+scores them against the user's resume, and returns a ranked table.
+
+DO NOT use search_jobs for this — search_jobs is for discovering new companies.
+openings_scan is for drilling into companies the user has already vetted.`,
+    parameters: {
+        type: Type.OBJECT,
+        properties: {
+            companies: {
+                type: Type.ARRAY,
+                items: { type: Type.STRING },
+                description: "Optional. Company names to scan (e.g. ['Vercel', 'LangChain']). Default: all Tier 1 companies with status 'To Apply' or 'Applied'.",
+            },
+            tier: {
+                type: Type.NUMBER,
+                description: "Optional. Filter by tier (1, 2, or 3). Default: 1.",
+            },
+            role_filter: {
+                type: Type.STRING,
+                description: "Optional. Keyword to filter job titles (e.g. 'engineer', 'solutions'). Case-insensitive.",
+            },
+            min_score: {
+                type: Type.NUMBER,
+                description: "Optional. Minimum resume match score to include (0–100). Default: 40.",
+            },
+            save: {
+                type: Type.BOOLEAN,
+                description: "Optional. Save discovered openings to job_openings.csv. Default: true.",
+            },
+            max_companies: {
+                type: Type.NUMBER,
+                description: "Optional. Max number of companies to scan per call. Default: 10 (to avoid long waits).",
+            },
+        },
+        required: [],
+    },
+    execute: async (args) => {
+        try {
+            // Load companies from jobs.csv
+            const jobsCsvPath = resolve(__dirname, "..", "..", "..", "..", "career-ops", "data", "jobs.csv");
+            const altPath = resolve(homedir(), ".careervivid", "jobs.csv");
+            const csvPath = existsSync(jobsCsvPath) ? jobsCsvPath : altPath;
+            if (!existsSync(csvPath)) {
+                return "❌ No jobs.csv found. Add some companies to your tracker first with tracker_add_job.";
+            }
+            const raw = readFileSync(csvPath, "utf-8");
+            const lines = raw.trim().split("\n");
+            const headers = lines[0].split(",");
+            const allJobs = lines.slice(1).filter(l => l.trim()).map(line => {
+                const cols = parseCsvLine(line);
+                const row = {};
+                headers.forEach((h, i) => { row[h] = (cols[i] ?? "").trim(); });
+                return row;
+            });
+            // Filter companies to scan
+            let candidates = allJobs;
+            if (args.companies && args.companies.length > 0) {
+                const names = args.companies.map(c => c.toLowerCase());
+                candidates = allJobs.filter(j => names.some(n => j.company?.toLowerCase().includes(n)));
+            }
+            else {
+                // Default: Tier 1 (or specified tier) with non-terminal status
+                const targetTier = String(args.tier ?? 1);
+                const TERMINAL_STATUSES = ["Rejected", "Ghosted", "Withdrawn", "Closed"];
+                candidates = allJobs.filter(j => j.tier === targetTier &&
+                    !TERMINAL_STATUSES.includes(j.status) &&
+                    j.careers_url?.startsWith("http"));
+            }
+            const maxCompanies = Math.min(args.max_companies ?? 10, 20);
+            candidates = candidates.slice(0, maxCompanies);
+            if (candidates.length === 0) {
+                return "ℹ️ No matching companies found in your tracker. Add companies with tracker_add_job first, or broaden the filter.";
+            }
+            const minScore = args.min_score ?? 40;
+            const keywords = await getResumeKeywords();
+            const scanResults = [];
+            process.stderr.write(`[openings_scan] Scanning ${candidates.length} companies...\n`);
+            // Fetch from each ATS
+            for (const job of candidates) {
+                const { id, company, careers_url: careersUrl, ats } = job;
+                const slug = extractSlug(careersUrl, ats);
+                let postings = [];
+                let error;
+                try {
+                    if (ats === "Greenhouse" && slug) {
+                        postings = await fetchGreenhouseJobs(slug);
+                    }
+                    else if (ats === "Ashby" && slug) {
+                        postings = await fetchAshbyJobs(slug);
+                    }
+                    else if (ats === "Lever" && slug) {
+                        postings = await fetchLeverJobs(slug, careersUrl);
+                    }
+                    else {
+                        // Direct or unknown ATS — try HTML parse
+                        postings = await fetchDirectJobs(careersUrl);
+                    }
+                }
+                catch (err) {
+                    error = err.message?.slice(0, 100);
+                }
+                scanResults.push({ company, companyId: id, ats, postings, error });
+                process.stderr.write(`  ${company}: ${postings.length} postings${error ? ` (⚠️ ${error})` : ""}\n`);
+            }
+            // Score + filter postings
+            const { rows: existingOpenings, path: openingsPath } = loadOpenings();
+            const newRows = [];
+            const reportSections = [];
+            for (const result of scanResults) {
+                const { company, companyId, ats, postings, error } = result;
+                if (error && postings.length === 0) {
+                    reportSections.push(`\n⚠️  ${company} (${ats}): Could not fetch — ${error}`);
+                    continue;
+                }
+                // Apply role filter
+                let filtered = postings;
+                if (args.role_filter) {
+                    const kw = args.role_filter.toLowerCase();
+                    filtered = postings.filter(p => p.title.toLowerCase().includes(kw));
+                }
+                // Score + sort
+                const scored = filtered
+                    .map(p => ({ posting: p, ...scorePosting(keywords, p) }))
+                    .filter(s => s.score >= minScore)
+                    .sort((a, b) => b.score - a.score);
+                if (scored.length === 0) {
+                    reportSections.push(`\n○ ${company} (${ats}) — ${postings.length} postings, none matched score ≥${minScore}%`);
+                    continue;
+                }
+                const lines = [
+                    `\n◆ ${company} (${ats}) — ${postings.length} open roles → ${scored.length} match above ${minScore}%`,
+                ];
+                for (const { posting, score, matched } of scored.slice(0, 8)) {
+                    lines.push(`  ★ [${score}%] ${posting.title}${posting.location ? `  [${posting.location}]` : ""}`, `     → ${posting.applyUrl}`);
+                    if (matched.length > 0) {
+                        lines.push(`     ✓ Keywords: ${matched.slice(0, 5).join(", ")}`);
+                    }
+                    // Build new row for job_openings.csv
+                    if (args.save !== false) {
+                        const existing = existingOpenings.find(r => r.company_id === companyId && r.posting_title === posting.title);
+                        if (!existing) {
+                            const newId = generateOpeningId(companyId, [...existingOpenings, ...newRows]);
+                            newRows.push({
+                                id: newId,
+                                company_id: companyId,
+                                company,
+                                posting_title: posting.title,
+                                apply_url: posting.applyUrl,
+                                ats,
+                                location: posting.location ?? "",
+                                match_score: String(score),
+                                match_keywords: matched.slice(0, 10).join("; "),
+                                status: "Found",
+                                date_found: today(),
+                                date_applied: "",
+                                notes: "",
+                            });
+                        }
+                    }
+                }
+                reportSections.push(lines.join("\n"));
+            }
+            // Persist new rows
+            let savedCount = 0;
+            if (args.save !== false && newRows.length > 0) {
+                const allRows = [...existingOpenings, ...newRows];
+                saveOpenings(allRows, openingsPath);
+                savedCount = newRows.length;
+            }
+            const totalFound = scanResults.reduce((s, r) => s + r.postings.length, 0);
+            const totalMatched = scanResults.reduce((s, r) => s + r.postings.filter(p => scorePosting(keywords, p).score >= minScore).length, 0);
+            return [
+                `🔍 Job Openings Scan — ${candidates.length} companies, ${totalFound} live postings, ${totalMatched} matches`,
+                "─".repeat(70),
+                ...reportSections,
+                "─".repeat(70),
+                savedCount > 0
+                    ? `\n✅ Saved ${savedCount} new openings to job_openings.csv. Say "show my openings" to list them.`
+                    : "\n(No new openings saved — all already tracked or save=false.)",
+                `\nTo apply: tell me which role interests you and I can help autofill the application.`,
+            ].join("\n");
+        }
+        catch (err) {
+            return `❌ openings_scan failed: ${err.message}`;
+        }
+    },
+};
+// ---------------------------------------------------------------------------
+// Tool: openings_list
+// ---------------------------------------------------------------------------
+export const OpeningsListTool = {
+    name: "openings_list",
+    description: `List saved job openings from job_openings.csv.
+Shows specific postings found by openings_scan, with match scores and direct apply URLs.
+Use this to review what's been found and decide what to apply to.
+
+Use when the user asks:
+- "Show my job openings"
+- "What openings have you found?"
+- "List the jobs I should apply to"
+- "What did the scan find?"`,
+    parameters: {
+        type: Type.OBJECT,
+        properties: {
+            status_filter: {
+                type: Type.STRING,
+                description: "Optional. Filter by status: Found | Reviewing | Applied | Rejected | Ghosted. Default: all.",
+            },
+            company_filter: {
+                type: Type.STRING,
+                description: "Optional. Filter by company name (partial match).",
+            },
+            min_score: {
+                type: Type.NUMBER,
+                description: "Optional. Minimum match score to show. Default: 0.",
+            },
+            sort_by: {
+                type: Type.STRING,
+                description: "Optional. Sort by: match_score | date_found | company. Default: match_score.",
+            },
+        },
+        required: [],
+    },
+    execute: async (args) => {
+        try {
+            const { rows } = loadOpenings();
+            if (rows.length === 0) {
+                return "ℹ️ No job openings saved yet. Run openings_scan to discover real job postings from your tracked companies.";
+            }
+            let filtered = rows;
+            if (args.status_filter) {
+                filtered = filtered.filter(r => r.status.toLowerCase() === args.status_filter.toLowerCase());
+            }
+            if (args.company_filter) {
+                const kw = args.company_filter.toLowerCase();
+                filtered = filtered.filter(r => r.company.toLowerCase().includes(kw));
+            }
+            if (args.min_score) {
+                filtered = filtered.filter(r => Number(r.match_score) >= args.min_score);
+            }
+            // Sort
+            const sortBy = args.sort_by ?? "match_score";
+            filtered.sort((a, b) => {
+                if (sortBy === "match_score")
+                    return Number(b.match_score) - Number(a.match_score);
+                if (sortBy === "date_found")
+                    return b.date_found.localeCompare(a.date_found);
+                if (sortBy === "company")
+                    return a.company.localeCompare(b.company);
+                return 0;
+            });
+            const lines = [
+                `📋 Job Openings — ${filtered.length} of ${rows.length} total`,
+                "─".repeat(70),
+            ];
+            // Group by status
+            const byStatus = new Map();
+            for (const r of filtered) {
+                const s = r.status || "Found";
+                if (!byStatus.has(s))
+                    byStatus.set(s, []);
+                byStatus.get(s).push(r);
+            }
+            const STATUS_ICONS = {
+                Found: "🔍", Reviewing: "📖", Applied: "✅", Rejected: "❌", Ghosted: "👻",
+            };
+            for (const [status, group] of byStatus) {
+                lines.push(`\n${STATUS_ICONS[status] ?? "○"} ${status} (${group.length})`);
+                for (const r of group) {
+                    lines.push(`  [${String(r.match_score).padStart(3)}%] ${r.company} — ${r.posting_title}`, `         📍 ${r.location || "Location not specified"}`, `         🔗 ${r.apply_url}`);
+                    if (r.date_applied)
+                        lines.push(`         Applied: ${r.date_applied}`);
+                }
+            }
+            lines.push("─".repeat(70));
+            lines.push(`\nTo apply: say "mark [opening-id] as applied on [date]" or "help me apply to [company] [role]".`);
+            lines.push(`Opening IDs: ${filtered.slice(0, 5).map(r => r.id).join(", ")}${filtered.length > 5 ? ", ..." : ""}`);
+            return lines.join("\n");
+        }
+        catch (err) {
+            return `❌ openings_list failed: ${err.message}`;
+        }
+    },
+};
+// ---------------------------------------------------------------------------
+// Tool: openings_apply
+// ---------------------------------------------------------------------------
+const OPENING_VALID_STATUSES = ["Found", "Reviewing", "Applied", "Rejected", "Ghosted"];
+export const OpeningsApplyTool = {
+    name: "openings_apply",
+    description: `Mark a specific job opening in job_openings.csv as Applied (or update its status).
+REQUIRES explicit date_applied when marking as Applied — never infer the date autonomously.
+
+Use when the user says:
+- "I applied to the Vercel Developer Success Engineer role today"
+- "Mark VER-OPN-001 as applied on 2026-04-14"
+- "Update the LangChain opening status to Reviewing"`,
+    parameters: {
+        type: Type.OBJECT,
+        properties: {
+            opening_id: {
+                type: Type.STRING,
+                description: "The opening ID from job_openings.csv (e.g. VER-OPN-001). Required.",
+            },
+            status: {
+                type: Type.STRING,
+                description: `New status: Found | Reviewing | Applied | Rejected | Ghosted.`,
+            },
+            date_applied: {
+                type: Type.STRING,
+                description: "ISO date of application (e.g. 2026-04-14). REQUIRED when status = Applied.",
+            },
+            notes: {
+                type: Type.STRING,
+                description: "Optional notes to append (not replace).",
+            },
+        },
+        required: ["opening_id"],
+    },
+    execute: async (args) => {
+        try {
+            const { rows, path } = loadOpenings();
+            const idx = rows.findIndex(r => r.id.toLowerCase() === args.opening_id.toLowerCase());
+            if (idx === -1) {
+                const ids = rows.slice(0, 10).map(r => r.id).join(", ");
+                return `❌ Opening ID "${args.opening_id}" not found.\nAvailable IDs: ${ids}`;
+            }
+            const row = { ...rows[idx] };
+            const changes = [];
+            if (args.status) {
+                if (!OPENING_VALID_STATUSES.includes(args.status)) {
+                    return `❌ Invalid status "${args.status}". Valid: ${OPENING_VALID_STATUSES.join(", ")}`;
+                }
+                // Gate: Applied requires explicit date
+                if (args.status === "Applied" && !args.date_applied && row.status !== "Applied") {
+                    return (`⚠️  CONFIRMATION REQUIRED\n` +
+                        `To mark "${row.company} — ${row.posting_title}" as Applied, provide the date:\n` +
+                        `  Example: "Yes, mark ${args.opening_id} as Applied on ${today()}"`);
+                }
+                changes.push(`status: ${row.status} → ${args.status}`);
+                row.status = args.status;
+            }
+            if (args.date_applied) {
+                row.date_applied = args.date_applied;
+                changes.push(`date_applied: ${args.date_applied}`);
+            }
+            else if (row.status === "Applied" && !row.date_applied) {
+                row.date_applied = today();
+                changes.push(`date_applied: ${today()} (auto)`);
+            }
+            if (args.notes) {
+                const existing = row.notes ? row.notes + "; " : "";
+                row.notes = (existing + args.notes).slice(0, 500);
+                changes.push(`notes: appended`);
+            }
+            rows[idx] = row;
+            saveOpenings(rows, path);
+            return [
+                `✅ Updated opening ${args.opening_id}:`,
+                ...changes.map(c => `  • ${c}`),
+                ``,
+                `${row.company} — ${row.posting_title}`,
+                `Status: ${row.status}${row.date_applied ? ` (applied ${row.date_applied})` : ""}`,
+                `🔗 ${row.apply_url}`,
+            ].join("\n");
+        }
+        catch (err) {
+            return `❌ openings_apply failed: ${err.message}`;
+        }
+    },
+};
+// ---------------------------------------------------------------------------
+// Export
+// ---------------------------------------------------------------------------
+export const ALL_JOB_OPENINGS_TOOLS = [
+    OpeningsScanTool,
+    OpeningsListTool,
+    OpeningsApplyTool,
+];

package/dist/commands/agent/repl.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"repl.d.ts","sourceRoot":"","sources":["../../../src/commands/agent/repl.ts"],"names":[],"mappings":"AAIA,OAAO,EAAE,sBAAsB,EAAE,MAAM,uCAAuC,CAAC;AAC/E,OAAO,EAAE,WAAW,EAAE,MAAM,4BAA4B,CAAC;AAEzD,OAAO,EAA4D,KAAK,WAAW,EAAE,MAAM,iBAAiB,CAAC;AAK7G,wBAAgB,iBAAiB,CAAC,SAAS,EAAE,MAAM,GAAG,IAAI,EAAE,KAAK,GAAE,MAAM,GAAG,IAAW,QAwBtF;AAED,wBAAsB,OAAO,CAC3B,MAAM,EAAE,WAAW,GAAG,sBAAsB,GAAG,IAAI,EACnD,OAAO,EAAE;IAAE,OAAO,CAAC,EAAE,OAAO,CAAC;IAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IAAC,UAAU,CAAC,EAAE,MAAM,CAAC;IAAC,OAAO,CAAC,EAAE,MAAM,CAAC;IAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IAAC,IAAI,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAA;CAAE,EAC9K,gBAAgB,EAAE,WAAW,EAC7B,aAAa,EAAE,MAAM,EACrB,QAAQ,EAAE,MAAM,GAAG,SAAS,EAC5B,iBAAiB,EAAE,MAAM,EACzB,KAAK,EAAE,GAAG,EAAE,GACX,OAAO,CAAC,IAAI,CAAC,CAmkBf"}
+{"version":3,"file":"repl.d.ts","sourceRoot":"","sources":["../../../src/commands/agent/repl.ts"],"names":[],"mappings":"AAIA,OAAO,EAAE,sBAAsB,EAAE,MAAM,uCAAuC,CAAC;AAC/E,OAAO,EAAE,WAAW,EAAE,MAAM,4BAA4B,CAAC;AAEzD,OAAO,EAA4D,KAAK,WAAW,EAAE,MAAM,iBAAiB,CAAC;AAK7G,wBAAgB,iBAAiB,CAAC,SAAS,EAAE,MAAM,GAAG,IAAI,EAAE,KAAK,GAAE,MAAM,GAAG,IAAW,QAwBtF;AAED,wBAAsB,OAAO,CAC3B,MAAM,EAAE,WAAW,GAAG,sBAAsB,GAAG,IAAI,EACnD,OAAO,EAAE;IAAE,OAAO,CAAC,EAAE,OAAO,CAAC;IAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IAAC,UAAU,CAAC,EAAE,MAAM,CAAC;IAAC,OAAO,CAAC,EAAE,MAAM,CAAC;IAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IAAC,IAAI,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAA;CAAE,EAC9K,gBAAgB,EAAE,WAAW,EAC7B,aAAa,EAAE,MAAM,EACrB,QAAQ,EAAE,MAAM,GAAG,SAAS,EAC5B,iBAAiB,EAAE,MAAM,EACzB,KAAK,EAAE,GAAG,EAAE,GACX,OAAO,CAAC,IAAI,CAAC,CAskBf"}

package/dist/commands/agent/repl.js

@@ -34,7 +34,7 @@ export async function askLoop(engine, options, selectedProvider, selectedModel,
     const WRITE_TOOLS = new Set([
         "tracker_add_job", "tracker_update_job", "kanban_add_job", "kanban_update_status",
         "save_cover_letter", "delete_cover_letter", "write_file", "patch_file",
-        "tracker_recheck_urls",
+        "tracker_recheck_urls", "openings_apply",
     ]);
     const SESSION_MAX_MUTATIONS = 25;
     const TURN_MAX_MUTATIONS = 10;
@@ -247,6 +247,9 @@ export async function askLoop(engine, options, selectedProvider, selectedModel,
                 verify_url: "🔍 Verifying URL...",
                 verify_job_urls: "🔍 Verifying job URLs...",
                 search_jobs: "🔍 Searching jobs...",
+                openings_scan: "🎯 Scanning companies for open roles...",
+                openings_list: "📋 Loading saved openings...",
+                openings_apply: "✅ Marking opening as applied...",
                 get_resume: "📄 Loading resume...",
                 list_resumes: "📄 Loading resumes...",
                 get_profile: "👤 Loading profile...",

package/dist/commands/agent/toolRegistry.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"toolRegistry.d.ts","sourceRoot":"","sources":["../../../src/commands/agent/toolRegistry.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,IAAI,EAAE,MAAM,qBAAqB,CAAC;AAuO3C,wBAAgB,QAAQ,CAAC,OAAO,EAAE;IAAE,IAAI,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAA;CAAE,GAAG,IAAI,EAAE,CA2ChG"}
+{"version":3,"file":"toolRegistry.d.ts","sourceRoot":"","sources":["../../../src/commands/agent/toolRegistry.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,IAAI,EAAE,MAAM,qBAAqB,CAAC;AAwO3C,wBAAgB,QAAQ,CAAC,OAAO,EAAE;IAAE,IAAI,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAA;CAAE,GAAG,IAAI,EAAE,CA8ChG"}

package/dist/commands/agent/toolRegistry.js

@@ -6,6 +6,7 @@ import { ALL_LOCAL_TRACKER_TOOLS } from "../../agent/tools/local-tracker.js";
 import { ALL_URL_VERIFIER_TOOLS } from "../../agent/tools/urlVerifier.js";
 import { ALL_PORTFOLIO_TOOLS } from "../../agent/tools/portfolio.js";
 import { ALL_COVERLETTER_TOOLS } from "../../agent/tools/coverLetter.js";
+import { ALL_JOB_OPENINGS_TOOLS } from "../../agent/tools/jobOpenings.js";
 import { publishSingleFile } from "../publish.js";
 // ── Publish tools ─────────────────────────────────────────────────────────────
 const PublishArticleTool = {
@@ -237,6 +238,10 @@ export function getTools(options) {
             if (!tools.find((x) => x.name === t.name))
                 tools.push(t);
         }
+        for (const t of ALL_JOB_OPENINGS_TOOLS) {
+            if (!tools.find((x) => x.name === t.name))
+                tools.push(t);
+        }
         return tools;
     }
     if (options.resume) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "careervivid",
-  "version": "1.12.36",
+  "version": "1.12.38",
   "description": "Official CLI for CareerVivid — publish articles, diagrams, and portfolio updates from your terminal or AI agent",
   "type": "module",
   "bin": {