Add 8 site adapters, early stop, Telegram notifications, and list command

New site adapters:
- usajobs: multi-keyword sweep (dentist/orthodontics/orthodontist), URL pagination
- docshealth, southernortho: Paylocity platform, keyword in URL
- pdshealth, saltdental: iCIMS Angular platform, URL pagination
- hospitaljobsonline: Load More button, relative URL fix
- aaoinfo: AAO Career Center, click-based AJAX pagination
- aroragroup: Load More + post-scrape keyword filter

Core improvements:
- Early stop pagination: stops when all fingerprints on a page are already in DB
- multi_keyword_mode: separate — one scrape per keyword, cross-sweep dedup
- gimme-job list command to view collected postings
- Stored column in status command
- Telegram notification (replacing KakaoTalk)
- DB repo: count_by_site(), get_existing_fingerprints()

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

gimme_job/adapters/aaoinfo.py

@@ -0,0 +1,59 @@
+"""AAO Career Center adapter — click-based AJAX pagination."""
+from __future__ import annotations
+
+from loguru import logger
+
+from gimme_job.adapters.base import ManifestDrivenAdapter
+from gimme_job.adapters.registry import register
+from gimme_job.models.dto import JobPostingCandidate, RawJobCard
+
+
+@register("aaoinfo")
+class AAOInfoAdapter(ManifestDrivenAdapter):
+    _BASE_URL = "https://careers.aaoinfo.org"
+
+    def _wait_for_results(self, page) -> None:
+        try:
+            page.wait_for_selector("div.job-tile", timeout=15000, state="attached")
+        except Exception:
+            logger.debug("[aaoinfo] Results wait timed out — may be zero results")
+
+    def collect_cards(self, page, manifest):
+        self._wait_for_results(page)
+        return super().collect_cards(page, manifest)
+
+    def paginate(self, page, page_index: int, config) -> bool:
+        """Click the next page number button in the AJAX pagination."""
+        if page_index >= config.pagination.max_pages - 1:
+            return False
+
+        next_page = page_index + 2
+
+        # Try clicking the specific page number button
+        next_btn = page.query_selector(f"#page-item-{next_page} a")
+        if not next_btn:
+            # Fall back to the last non-active/non-disabled pagination link
+            next_btn = page.query_selector(
+                "ul.pagination li.page-item:not(.active):not(.disabled):last-child a"
+            )
+
+        if not next_btn:
+            logger.debug(f"[aaoinfo] No next page button found — stopping at page {page_index + 1}")
+            return False
+
+        logger.debug(f"[aaoinfo] Paginating to page {next_page}")
+        next_btn.click()
+
+        # Wait for the new page's active indicator to confirm AJAX completed
+        try:
+            page.wait_for_selector(f"#page-item-{next_page}.active", timeout=12000)
+        except Exception:
+            page.wait_for_timeout(3000)
+
+        return True
+
+    def normalize(self, raw: RawJobCard) -> JobPostingCandidate:
+        candidate = super().normalize(raw)
+        if candidate.job_url and candidate.job_url.startswith("/"):
+            candidate.job_url = self._BASE_URL + candidate.job_url
+        return candidate

gimme_job/adapters/aroragroup.py

@@ -0,0 +1,68 @@
+"""Arora Group Jobs adapter — Load More button, keyword post-filter."""
+from __future__ import annotations
+
+from loguru import logger
+
+from gimme_job.adapters.base import ManifestDrivenAdapter
+from gimme_job.adapters.registry import register
+from gimme_job.models.dto import JobPostingCandidate, RawJobCard
+
+_KEYWORDS = ["dentist", "orthodontic", "orthodontist"]
+_MAX_LOAD_MORE = 3  # clicks in addition to the initial load
+
+
+@register("aroragroup")
+class AroraGroupAdapter(ManifestDrivenAdapter):
+    _BASE_URL = "https://jobs.aroragroup.com"
+
+    def collect_cards(self, page, manifest):
+        """Load initial results, click 'Load More' up to N times, then filter by keywords."""
+        try:
+            page.wait_for_selector(".job-post-row", timeout=15000, state="attached")
+        except Exception:
+            logger.debug("[aroragroup] Initial results wait timed out")
+
+        for i in range(_MAX_LOAD_MORE):
+            btn = page.query_selector("button#loadMore")
+            if not btn or not btn.is_visible():
+                logger.debug(f"[aroragroup] Load More button gone after {i} extra loads")
+                break
+
+            current_count = len(page.query_selector_all(".job-post-row[onclick]"))
+            btn.click()
+
+            try:
+                page.wait_for_function(
+                    f"document.querySelectorAll('.job-post-row[onclick]').length > {current_count}",
+                    timeout=15000,
+                )
+                logger.debug(f"[aroragroup] Load More {i + 1}/{_MAX_LOAD_MORE}: loaded more cards")
+            except Exception:
+                logger.debug(f"[aroragroup] Load More {i + 1}: no new cards appeared — stopping")
+                break
+
+        all_cards = super().collect_cards(page, manifest)
+
+        # Filter by keywords in title
+        filtered = [c for c in all_cards if self._matches_keywords(c.title)]
+        logger.debug(
+            f"[aroragroup] Keyword filter: {len(filtered)}/{len(all_cards)} cards matched"
+        )
+        return filtered
+
+    def paginate(self, page, page_index: int, config) -> bool:
+        """Pagination is handled inside collect_cards — always return False."""
+        return False
+
+    def normalize(self, raw: RawJobCard) -> JobPostingCandidate:
+        candidate = super().normalize(raw)
+        if candidate.job_url and candidate.job_url.startswith("/"):
+            candidate.job_url = self._BASE_URL + candidate.job_url
+        return candidate
+
+    @staticmethod
+    def _matches_keywords(title: str | None) -> bool:
+        if not title:
+            return False
+        title_lower = title.lower()
+        return any(kw in title_lower for kw in _KEYWORDS)

gimme_job/adapters/docshealth.py

@@ -0,0 +1,34 @@
+"""DOCS Health adapter — Paylocity-hosted job board, single-page results."""
+from __future__ import annotations
+
+from loguru import logger
+
+from gimme_job.adapters.base import ManifestDrivenAdapter
+from gimme_job.adapters.registry import register
+from gimme_job.models.dto import JobPostingCandidate, RawJobCard
+
+
+@register("docshealth")
+class DocsHealthAdapter(ManifestDrivenAdapter):
+    _BASE_URL = "https://recruiting.paylocity.com"
+    _COMPANY = "DOCS Health"
+
+    def collect_cards(self, page, manifest):
+        # Wait for job listings to render
+        try:
+            page.wait_for_selector(".job-listing-job-item", timeout=12000, state="attached")
+        except Exception:
+            logger.debug("[docshealth] Results wait timed out — may be zero results")
+        return super().collect_cards(page, manifest)
+
+    def normalize(self, raw: RawJobCard) -> JobPostingCandidate:
+        candidate = super().normalize(raw)
+        # Fix relative URL → absolute
+        if candidate.job_url and candidate.job_url.startswith("/"):
+            candidate.job_url = self._BASE_URL + candidate.job_url
+        # Company is always DOCS Health (not in per-card HTML)
+        candidate.company = self._COMPANY
+        # Clean posted_text: "03/27/2026 - " → "03/27/2026"
+        if candidate.posted_text:
+            candidate.posted_text = candidate.posted_text.replace(" - ", "").strip()
+        return candidate

gimme_job/adapters/hospitaljobsonline.py

@@ -0,0 +1,65 @@
+"""Hospital Jobs Online adapter — URL-based pagination with page= parameter."""
+from __future__ import annotations
+
+import re
+
+from loguru import logger
+
+from gimme_job.adapters.base import ManifestDrivenAdapter
+from gimme_job.adapters.registry import register
+from gimme_job.models.dto import JobPostingCandidate, RawJobCard
+
+
+@register("hospitaljobsonline")
+class HospitalJobsOnlineAdapter(ManifestDrivenAdapter):
+    _BASE_URL = "https://www.hospitaljobsonline.com"
+
+    def _wait_for_results(self, page) -> None:
+        try:
+            page.wait_for_selector("div.jobresult", timeout=15000, state="attached")
+        except Exception:
+            logger.debug("[hospitaljobsonline] Results wait timed out — may be zero results")
+
+    def collect_cards(self, page, manifest):
+        self._wait_for_results(page)
+        return super().collect_cards(page, manifest)
+
+    def paginate(self, page, page_index: int, config) -> bool:
+        """Navigate to next page by incrementing the ?page= URL parameter."""
+        if page_index >= config.pagination.max_pages - 1:
+            return False
+
+        current_url = page.url
+        next_page = page_index + 2
+
+        if re.search(r"[?&]page=\d+", current_url):
+            next_url = re.sub(r"(page=)\d+", f"page={next_page}", current_url)
+        else:
+            sep = "&" if "?" in current_url else "?"
+            next_url = f"{current_url}{sep}page={next_page}"
+
+        logger.debug(f"[hospitaljobsonline] Paginating to page {next_page}: {next_url}")
+        page.goto(next_url, timeout=30000)
+        self._wait_for_results(page)
+        return True
+
+    def normalize(self, raw: RawJobCard) -> JobPostingCandidate:
+        candidate = super().normalize(raw)
+
+        if candidate.job_url and candidate.job_url.startswith("/"):
+            candidate.job_url = self._BASE_URL + candidate.job_url
+
+        # Parse location text: "1 week ago - Company Name - Location: City, ST"
+        if candidate.location:
+            loc = candidate.location
+            # Extract part after "Location:" if present
+            m = re.search(r"Location:\s*(.+)$", loc, re.IGNORECASE)
+            if m:
+                candidate.location = m.group(1).strip()
+            else:
+                # Fallback: take last segment after " - "
+                parts = [p.strip() for p in loc.split(" - ")]
+                if len(parts) >= 2:
+                    candidate.location = parts[-1]
+
+        return candidate

gimme_job/adapters/pdshealth.py

@@ -0,0 +1,48 @@
+"""PDS Health adapter — Angular Material job board, URL-based pagination."""
+from __future__ import annotations
+
+import re
+
+from loguru import logger
+
+from gimme_job.adapters.base import ManifestDrivenAdapter
+from gimme_job.adapters.registry import register
+from gimme_job.models.dto import JobPostingCandidate, RawJobCard
+
+
+@register("pdshealth")
+class PdsHealthAdapter(ManifestDrivenAdapter):
+
+    def _wait_for_results(self, page) -> None:
+        try:
+            page.wait_for_selector(
+                "mat-expansion-panel.search-result-item",
+                timeout=15000,
+                state="attached",
+            )
+        except Exception:
+            logger.debug("[pdshealth] Results wait timed out — may be zero results")
+
+    def collect_cards(self, page, manifest):
+        self._wait_for_results(page)
+        return super().collect_cards(page, manifest)
+
+    def paginate(self, page, page_index: int, config) -> bool:
+        """Navigate to next page by incrementing the &page= URL parameter."""
+        max_pages = config.pagination.max_pages
+        if page_index >= max_pages - 1:
+            return False
+
+        current_url = page.url
+        next_page = page_index + 2  # 0-based index → 1-based page param
+
+        if re.search(r"[?&]page=\d+", current_url):
+            next_url = re.sub(r"(page=)\d+", f"page={next_page}", current_url)
+        else:
+            sep = "&" if "?" in current_url else "?"
+            next_url = f"{current_url}{sep}page={next_page}"
+
+        logger.debug(f"[pdshealth] Paginating to page {next_page}: {next_url}")
+        page.goto(next_url, timeout=30000)
+        self._wait_for_results(page)
+        return True

gimme_job/adapters/saltdental.py

@@ -0,0 +1,46 @@
+"""Salt Dental Partners adapter — iCIMS Angular platform, URL-based pagination."""
+from __future__ import annotations
+
+import re
+
+from loguru import logger
+
+from gimme_job.adapters.base import ManifestDrivenAdapter
+from gimme_job.adapters.registry import register
+
+
+@register("saltdental")
+class SaltDentalAdapter(ManifestDrivenAdapter):
+
+    def _wait_for_results(self, page) -> None:
+        try:
+            page.wait_for_selector(
+                "mat-expansion-panel.search-result-item",
+                timeout=15000,
+                state="attached",
+            )
+        except Exception:
+            logger.debug("[saltdental] Results wait timed out — may be zero results")
+
+    def collect_cards(self, page, manifest):
+        self._wait_for_results(page)
+        return super().collect_cards(page, manifest)
+
+    def paginate(self, page, page_index: int, config) -> bool:
+        """Navigate to next page by incrementing the &page= URL parameter."""
+        if page_index >= config.pagination.max_pages - 1:
+            return False
+
+        current_url = page.url
+        next_page = page_index + 2
+
+        if re.search(r"[?&]page=\d+", current_url):
+            next_url = re.sub(r"(page=)\d+", f"page={next_page}", current_url)
+        else:
+            sep = "&" if "?" in current_url else "?"
+            next_url = f"{current_url}{sep}page={next_page}"
+
+        logger.debug(f"[saltdental] Paginating to page {next_page}: {next_url}")
+        page.goto(next_url, timeout=30000)
+        self._wait_for_results(page)
+        return True

gimme_job/adapters/southernortho.py

@@ -0,0 +1,28 @@
+"""Southern Orthodontic Partners adapter — Paylocity platform, single page."""
+from __future__ import annotations
+
+from loguru import logger
+
+from gimme_job.adapters.base import ManifestDrivenAdapter
+from gimme_job.adapters.registry import register
+from gimme_job.models.dto import JobPostingCandidate, RawJobCard
+
+
+@register("southernortho")
+class SouthernOrthoAdapter(ManifestDrivenAdapter):
+    _BASE_URL = "https://recruiting.paylocity.com"
+
+    def collect_cards(self, page, manifest):
+        try:
+            page.wait_for_selector(".job-listing-job-item", timeout=12000, state="attached")
+        except Exception:
+            logger.debug("[southernortho] Results wait timed out — may be zero results")
+        return super().collect_cards(page, manifest)
+
+    def normalize(self, raw: RawJobCard) -> JobPostingCandidate:
+        candidate = super().normalize(raw)
+        if candidate.job_url and candidate.job_url.startswith("/"):
+            candidate.job_url = self._BASE_URL + candidate.job_url
+        if candidate.posted_text:
+            candidate.posted_text = candidate.posted_text.replace(" - ", "").strip()
+        return candidate

gimme_job/adapters/usajobs.py

@@ -0,0 +1,58 @@
+"""USAJOBS adapter — waits for JS render, handles relative URLs, URL-based pagination."""
+from __future__ import annotations
+
+import re
+
+from loguru import logger
+
+from gimme_job.adapters.base import ManifestDrivenAdapter
+from gimme_job.adapters.registry import register
+from gimme_job.models.dto import JobPostingCandidate, RawJobCard
+
+
+@register("usajobs")
+class UsaJobsAdapter(ManifestDrivenAdapter):
+    _BASE_URL = "https://www.usajobs.gov"
+
+    def _wait_for_results(self, page) -> None:
+        """Wait for JS-rendered results or zero-results heading."""
+        try:
+            page.wait_for_selector(
+                "#search-results .border, h2.font-normal",
+                timeout=12000,
+                state="attached",
+            )
+        except Exception:
+            logger.debug("[usajobs] Results wait timed out — may be zero results")
+
+    def collect_cards(self, page, manifest):
+        self._wait_for_results(page)
+        return super().collect_cards(page, manifest)
+
+    def paginate(self, page, page_index: int, config) -> bool:
+        """Navigate to next page by incrementing the &p= URL parameter."""
+        max_pages = config.pagination.max_pages
+        if page_index >= max_pages - 1:
+            return False
+
+        current_url = page.url
+        next_page = page_index + 2  # page_index is 0-based; URL param is 1-based
+
+        # Replace &p=N or add &p=N
+        if re.search(r"[?&]p=\d+", current_url):
+            next_url = re.sub(r"(p=)\d+", f"p={next_page}", current_url)
+        else:
+            sep = "&" if "?" in current_url else "?"
+            next_url = f"{current_url}{sep}p={next_page}"
+
+        logger.debug(f"[usajobs] Paginating to page {next_page}: {next_url}")
+        page.goto(next_url, timeout=30000)
+        self._wait_for_results(page)
+        return True
+
+    def normalize(self, raw: RawJobCard) -> JobPostingCandidate:
+        candidate = super().normalize(raw)
+        # USAJOBS job links are relative: /job/123456 → absolute URL
+        if candidate.job_url and candidate.job_url.startswith("/"):
+            candidate.job_url = self._BASE_URL + candidate.job_url
+        return candidate

gimme_job/cli.py

@@ -94,13 +94,14 @@ def _write_default_global_yaml(path: Path) -> None:
             "max_input_items": 200,
         },
         "notification": {
-            "provider": "kakaotalk",
+            "provider": "telegram",
             "fallback_markdown": True,
         },
     }
     write_yaml(path, data)
 
 
+
 def _check_playwright() -> None:
     try:
         from playwright.sync_api import sync_playwright
@@ -141,6 +142,44 @@ def _check_claude() -> None:
         console.print("[yellow]![/yellow] Claude Code CLI not found (needed for learn/repair)")
 
 
+# ── login ─────────────────────────────────────────────────────────────────────
+
+
+@app.command()
+def login() -> None:
+    """Open the JobAgent browser to log in to sites manually.
+
+    Browse to any sites (LinkedIn, Indeed, Google, etc.), log in, then press Enter.
+    All cookies and localStorage are saved in the JobAgent profile and reused on
+    every subsequent run.
+    """
+    from gimme_job.config import load_global_config
+    from gimme_job.runtime.browser import BrowserManager
+
+    cfg = load_global_config()
+
+    console.print(Panel("[bold cyan]gimme-job login[/bold cyan]", expand=False))
+    console.print(
+        "Opening JobAgent Chrome profile.\n"
+        "Log in to any sites you need, then press Enter here to close the browser.\n"
+        "All cookies and localStorage will be saved and reused on future runs.\n"
+    )
+
+    bm = BrowserManager(
+        profile_name=cfg.runtime.profile_name,
+        headless=False,
+        slow_mo=cfg.runtime.slow_mo_ms,
+    )
+    try:
+        bm.open_context()
+        bm.new_page()
+        input("  >> Press Enter when done... ")
+    finally:
+        bm.close()
+
+    console.print("[green]✓[/green] Browser closed. Session saved to JobAgent profile.")
+
+
 # ── run ───────────────────────────────────────────────────────────────────────
 
 
@@ -148,7 +187,7 @@ def _check_claude() -> None:
 def run(
     site: Optional[str] = typer.Option(None, "--site", help="Run a single site only"),
     dry_run: bool = typer.Option(False, "--dry-run", help="Extract but do not save or notify"),
-    skip_notify: bool = typer.Option(False, "--skip-notify", help="Skip KakaoTalk notification"),
+    skip_notify: bool = typer.Option(False, "--skip-notify", help="Skip Telegram notification"),
     today_only: bool = typer.Option(False, "--today-only", help="Only process today's new items"),
     limit_sites: Optional[int] = typer.Option(None, "--limit-sites", help="Max sites to run"),
 ) -> None:
@@ -290,7 +329,7 @@ def test(
 def notify(
     today: bool = typer.Option(True, "--today/--no-today", help="Re-send today's digest"),
 ) -> None:
-    """Re-send today's job summary via KakaoTalk."""
+    """Re-send today's job summary via Telegram."""
     from datetime import date
 
     from gimme_job.config import load_global_config
@@ -331,6 +370,59 @@ def notify(
         console.print("[yellow]![/yellow] Notification failed — saved as markdown fallback")
 
 
+# ── list ──────────────────────────────────────────────────────────────────────
+
+
+@app.command(name="list")
+def list_postings(
+    site: Optional[str] = typer.Option(None, "--site", help="Filter by site ID"),
+    today: bool = typer.Option(False, "--today", help="Only show today's new postings"),
+    limit: int = typer.Option(50, "--limit", help="Max rows to show"),
+) -> None:
+    """List collected job postings from the database."""
+    from datetime import date
+
+    from gimme_job.db.engine import get_engine, get_session_factory
+    from gimme_job.db.repo import JobPostingRepo
+    from gimme_job.models.db import JobPosting
+    from sqlalchemy import select
+
+    engine = get_engine()
+    factory = get_session_factory(engine)
+
+    table = Table(title="Job Postings", show_header=True, show_lines=False)
+    table.add_column("#", style="dim", justify="right", width=4)
+    table.add_column("Site", style="cyan", width=10)
+    table.add_column("Title", width=40)
+    table.add_column("Company", width=25)
+    table.add_column("Location", width=20)
+    table.add_column("Posted", width=12)
+    table.add_column("New", width=4)
+
+    with factory() as session:
+        q = select(JobPosting).order_by(JobPosting.first_seen_at.desc())
+        if site:
+            q = q.where(JobPosting.site_id == site)
+        if today:
+            q = q.where(JobPosting.run_date == date.today(), JobPosting.is_new == True)
+        q = q.limit(limit)
+        rows = list(session.execute(q).scalars().all())
+
+    for i, p in enumerate(rows, 1):
+        table.add_row(
+            str(i),
+            p.site_id,
+            p.title[:38] if p.title else "",
+            (p.company or "")[:23],
+            (p.location or "")[:18],
+            p.posted_text or "",
+            "[green]Y[/green]" if p.is_new else "",
+        )
+
+    console.print(table)
+    console.print(f"[dim]{len(rows)} rows (limit={limit})[/dim]")
+
+
 # ── status ────────────────────────────────────────────────────────────────────
 
 
@@ -344,10 +436,14 @@ def status() -> None:
     engine = get_engine()
     factory = get_session_factory(engine)
 
+    from sqlalchemy import func, select
+    from gimme_job.models.db import JobPosting
+
     table = Table(title="Site Status", show_header=True)
     table.add_column("Site", style="cyan")
     table.add_column("Enabled")
     table.add_column("Repair?")
+    table.add_column("Stored", justify="right")
     table.add_column("Last Run")
     table.add_column("Last Status")
     table.add_column("Failures", justify="right")
@@ -364,6 +460,9 @@ def status() -> None:
 
             config = session.get(__import__("gimme_job.models.db", fromlist=["SiteConfig"]).SiteConfig, site_id)
             recent = run_repo.get_recent_runs(session, site_id, limit=1)
+            stored = session.execute(
+                select(func.count()).where(JobPosting.site_id == site_id)
+            ).scalar() or 0
 
             enabled = "[green]yes[/green]" if manifest.enabled else "[dim]no[/dim]"
             repair = "[red]YES[/red]" if manifest.repair_needed else "[green]no[/green]"
@@ -371,7 +470,7 @@ def status() -> None:
             last_status = recent[0].status if recent else "[dim]-[/dim]"
             failures = str(config.consecutive_failures) if config else "0"
 
-            table.add_row(site_id, enabled, repair, last_run, last_status, failures)
+            table.add_row(site_id, enabled, repair, str(stored), last_run, last_status, failures)
 
     console.print(table)
 

gimme_job/config.py

@@ -40,7 +40,7 @@ class SummarizationConfig(BaseModel):
 
 
 class NotificationConfig(BaseModel):
-    provider: str = "kakaotalk"
+    provider: str = "telegram"
     fallback_markdown: bool = True
 
 

gimme_job/db/repo.py

@@ -88,6 +88,24 @@ class JobPostingRepo:
             ).scalars().all()
         )
 
+    def count_by_site(self, session: Session, site_id: str) -> int:
+        """Count total job postings stored for a site."""
+        from sqlalchemy import func
+        return session.execute(
+            select(func.count()).where(JobPosting.site_id == site_id)
+        ).scalar() or 0
+
+    def get_existing_fingerprints(
+        self, session: Session, fingerprints: list[str]
+    ) -> set[str]:
+        """Return the subset of given fingerprints that already exist in the DB."""
+        if not fingerprints:
+            return set()
+        rows = session.execute(
+            select(JobPosting.fingerprint).where(JobPosting.fingerprint.in_(fingerprints))
+        ).scalars().all()
+        return set(rows)
+
 
 class SiteRunRepo:
     def record_run(self, session: Session, result: SiteRunResult) -> SiteRun:

gimme_job/models/manifest.py

@@ -70,6 +70,7 @@ class SearchOverride(BaseModel):
     date_mode: Optional[str] = None
     sort: Optional[str] = None
     max_items: Optional[int] = None
+    multi_keyword_mode: Optional[str] = None  # "separate" → one scrape per keyword
 
 
 class LearnMeta(BaseModel):

gimme_job/runtime/browser.py

@@ -10,7 +10,12 @@ from loguru import logger
 class BrowserManager:
     """Manages Playwright browser contexts using persistent Chrome profiles."""
 
-    def __init__(self, profile_name: str = "JobAgent", headless: bool = True, slow_mo: int = 0):
+    def __init__(
+        self,
+        profile_name: str = "JobAgent",
+        headless: bool = True,
+        slow_mo: int = 0,
+    ):
         self.profile_name = profile_name
         self.headless = headless
         self.slow_mo = slow_mo
@@ -34,7 +39,6 @@ class BrowserManager:
 
         profile_path = chrome_profile_dir(name)
         profile_path.mkdir(parents=True, exist_ok=True)
-
         logger.info(f"Opening browser context: profile={name}, headless={hl}")
 
         self._playwright = sync_playwright().start()
@@ -46,12 +50,41 @@ class BrowserManager:
             args=[
                 "--disable-blink-features=AutomationControlled",
                 "--no-sandbox",
+                "--disable-infobars",
+                "--disable-dev-shm-usage",
             ],
+            ignore_default_args=["--enable-automation"],
             viewport={"width": 1280, "height": 900},
             locale="en-US",
+            user_agent=(
+                "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) "
+                "AppleWebKit/537.36 (KHTML, like Gecko) "
+                "Chrome/134.0.0.0 Safari/537.36"
+            ),
         )
+        # Apply stealth patches to every new page
+        try:
+            from playwright_stealth import stealth_sync
+            self._context.add_init_script("""
+                Object.defineProperty(navigator, 'webdriver', { get: () => undefined });
+            """)
+            # stealth_sync works per-page; store reference for use in open_page()
+            self._stealth_fn = stealth_sync
+        except ImportError:
+            self._stealth_fn = None
+
         return self._context
 
+    def new_page(self):
+        """Open a new page with stealth applied."""
+        page = self._context.new_page()
+        if getattr(self, "_stealth_fn", None):
+            try:
+                self._stealth_fn(page)
+            except Exception as e:
+                logger.debug(f"Stealth apply failed: {e}")
+        return page
+
     def close(self) -> None:
         if self._context:
             try:

gimme_job/runtime/notifier.py

@@ -1,4 +1,4 @@
-"""Notification dispatcher: KakaoTalk with Markdown fallback."""
+"""Notification dispatcher: Telegram with Markdown fallback."""
 from __future__ import annotations
 
 from datetime import date
@@ -15,25 +15,23 @@ class NotificationDispatcher:
 
     def send(self, summary: str, run_date: date) -> bool:
         """Send the summary. Returns True if any method succeeded."""
-        from gimme_job.db.repo import NotificationLogRepo
-        from gimme_job.runtime.kakao import KakaoTalkClient
+        from gimme_job.runtime.telegram import TelegramClient
 
-        client = KakaoTalkClient()
+        client = TelegramClient()
         success = False
 
         if client.is_configured():
-            ok = client.send_self_memo(summary)
+            ok = client.send_message(summary)
             if ok:
                 success = True
-                self._log_notification(run_date, "kakaotalk", "success")
+                self._log_notification(run_date, "telegram", "success")
             else:
-                logger.warning("KakaoTalk failed — writing markdown fallback")
-                self._log_notification(run_date, "kakaotalk", "failed", "send_self_memo returned False")
+                logger.warning("Telegram failed — writing markdown fallback")
+                self._log_notification(run_date, "telegram", "failed", "send_message returned False")
         else:
-            logger.warning("KakaoTalk not configured — writing markdown fallback only")
-            self._log_notification(run_date, "kakaotalk", "skipped", "not configured")
+            logger.warning("Telegram not configured — writing markdown fallback only")
+            self._log_notification(run_date, "telegram", "skipped", "not configured")
 
-        # Always save markdown fallback if configured
         if self.cfg.notification.fallback_markdown:
             path = self.send_markdown_fallback(summary, run_date)
             logger.info(f"Markdown saved: {path}")

gimme_job/runtime/orchestrator.py

@@ -56,9 +56,10 @@ class RunOrchestrator:
 
             # Record to DB (unless dry run)
             if not dry_run:
-                with self.session_factory() as session:
-                    from gimme_job.db.repo import SiteRunRepo
-                    from gimme_job.runtime.health import update_site_status
+                from gimme_job.db.engine import db_session
+                from gimme_job.db.repo import SiteRunRepo
+                from gimme_job.runtime.health import update_site_status
+                with db_session() as session:
                     SiteRunRepo().record_run(session, result)
                     update_site_status(site_id, result, session)
 
@@ -97,7 +98,21 @@ class RunOrchestrator:
         from gimme_job.utils.paths import screenshot_path, dom_snapshot_path, trace_path
 
         try:
-            candidates = self._scrape_site(site_id, manifest, query, run_id, result)
+            if manifest.search_override.multi_keyword_mode == "separate" and query.keywords:
+                # Run one scrape per keyword and merge results
+                from gimme_job.runtime.dedupe import deduplicate_in_batch
+                all_candidates = []
+                for kw in query.keywords:
+                    from gimme_job.models.dto import SearchQuery
+                    kw_query = query.model_copy(update={"keywords": [kw]})
+                    kw_run_id = f"{run_id}-{kw[:8]}"
+                    logger.info(f"[{site_id}] Keyword sweep: '{kw}'")
+                    kw_candidates = self._scrape_site(site_id, manifest, kw_query, kw_run_id, result, dry_run=dry_run)
+                    all_candidates.extend(kw_candidates)
+                # Deduplicate across keyword sweeps by fingerprint
+                candidates = deduplicate_in_batch(all_candidates)
+            else:
+                candidates = self._scrape_site(site_id, manifest, query, run_id, result, dry_run=dry_run)
         except Exception as e:
             logger.error(f"[{site_id}] Scrape error: {e}")
             result.status = RunStatus.FAILED
@@ -108,8 +123,9 @@ class RunOrchestrator:
         result.items_found = len(candidates)
 
         if not dry_run and candidates:
-            with self.session_factory() as session:
-                from gimme_job.db.repo import JobPostingRepo
+            from gimme_job.db.engine import db_session
+            from gimme_job.db.repo import JobPostingRepo
+            with db_session() as session:
                 total, new_count = JobPostingRepo().upsert_candidates(session, candidates, run_date)
                 result.new_items = new_count
         elif dry_run:
@@ -131,6 +147,7 @@ class RunOrchestrator:
         query: SearchQuery,
         run_id: str,
         result: SiteRunResult,
+        dry_run: bool = False,
     ):
         from gimme_job.adapters.registry import get_adapter
         from gimme_job.runtime.browser import BrowserManager
@@ -141,13 +158,17 @@ class RunOrchestrator:
         headless = manifest.browser.headless_on_run
         profile = manifest.browser.profile_name or self.cfg.runtime.profile_name
 
-        bm = BrowserManager(profile_name=profile, headless=headless, slow_mo=self.cfg.runtime.slow_mo_ms)
+        bm = BrowserManager(
+            profile_name=profile,
+            headless=headless,
+            slow_mo=self.cfg.runtime.slow_mo_ms,
+        )
         all_cards = []
 
         try:
             context = bm.open_context()
             bm.start_tracing()
-            page = context.new_page()
+            page = bm.new_page()
 
             # Set timeouts
             page.set_default_timeout(self.cfg.runtime.default_timeout_ms)
@@ -177,6 +198,17 @@ class RunOrchestrator:
                 except Exception:
                     logger.warning(f"[{site_id}] Wait selector timed out — continuing anyway")
 
+            # Check if site has existing data (for early stop on update runs)
+            site_has_data = False
+            if not dry_run:
+                try:
+                    from gimme_job.db.engine import db_session
+                    from gimme_job.db.repo import JobPostingRepo
+                    with db_session() as _s:
+                        site_has_data = JobPostingRepo().count_by_site(_s, site_id) > 0
+                except Exception:
+                    pass
+
             # Collect cards across pages
             max_pages = min(
                 manifest.pagination.max_pages,
@@ -187,6 +219,35 @@ class RunOrchestrator:
                 all_cards.extend(page_cards)
                 logger.debug(f"[{site_id}] Page {page_idx+1}: {len(page_cards)} cards")
 
+                if not page_cards:
+                    break  # no results on this page — stop paginating
+
+                # Early stop: if site has existing data and this page's cards are all known,
+                # we've caught up to previously stored content — no need to paginate further.
+                if site_has_data and page_idx >= 1:
+                    try:
+                        from gimme_job.db.engine import db_session
+                        from gimme_job.db.repo import JobPostingRepo
+                        page_fps = []
+                        for card in page_cards:
+                            try:
+                                c = adapter.normalize(card)
+                                if c.fingerprint:
+                                    page_fps.append(c.fingerprint)
+                            except Exception:
+                                pass
+                        if page_fps:
+                            with db_session() as _s:
+                                known = JobPostingRepo().get_existing_fingerprints(_s, page_fps)
+                            if len(known) >= len(page_fps):
+                                logger.info(
+                                    f"[{site_id}] Page {page_idx+1}: all {len(page_fps)} items"
+                                    " already in DB — stopping early"
+                                )
+                                break
+                    except Exception as e:
+                        logger.debug(f"[{site_id}] Early stop check failed: {e}")
+
                 if page_idx < max_pages - 1:
                     if not adapter.paginate(page, page_idx, manifest):
                         break
@@ -247,22 +308,25 @@ class RunOrchestrator:
 
     def _summarize_and_notify(self, summary: RunSummary, run_date: date) -> None:
         try:
-            with self.session_factory() as session:
-                from gimme_job.db.repo import JobPostingRepo, SummaryRepo
+            from gimme_job.db.engine import db_session
+            from gimme_job.db.repo import JobPostingRepo, SummaryRepo
+
+            with db_session() as session:
                 postings = JobPostingRepo().get_today_new(session, run_date)
 
-                if not postings:
-                    return
+            if not postings:
+                return
 
-                from gimme_job.runtime.summarizer import OllamaSummarizer
-                summarizer = OllamaSummarizer(
-                    base_url=self.cfg.summarization.ollama_base_url,
-                    model=self.cfg.summarization.model,
-                    temperature=self.cfg.summarization.temperature,
-                )
-                text = summarizer.summarize(postings)
-                summary.summary_text = text
+            from gimme_job.runtime.summarizer import OllamaSummarizer
+            summarizer = OllamaSummarizer(
+                base_url=self.cfg.summarization.ollama_base_url,
+                model=self.cfg.summarization.model,
+                temperature=self.cfg.summarization.temperature,
+            )
+            text = summarizer.summarize(postings)
+            summary.summary_text = text
 
+            with db_session() as session:
                 SummaryRepo().save_summary(session, run_date, text, self.cfg.summarization.model)
 
             from gimme_job.runtime.notifier import NotificationDispatcher

gimme_job/runtime/telegram.py

@@ -0,0 +1,53 @@
+"""Telegram notification client — sends messages via Bot API."""
+from __future__ import annotations
+
+import os
+from typing import Optional
+
+from loguru import logger
+
+TELEGRAM_API_URL = "https://api.telegram.org/bot{token}/sendMessage"
+MAX_TEXT_LENGTH = 4096  # Telegram message limit
+
+
+class TelegramClient:
+    def __init__(
+        self,
+        bot_token: Optional[str] = None,
+        chat_id: Optional[str] = None,
+    ):
+        self.bot_token = bot_token or os.environ.get("TELEGRAM_BOT_TOKEN", "")
+        self.chat_id = chat_id or os.environ.get("TELEGRAM_CHAT_ID", "")
+
+    def is_configured(self) -> bool:
+        return bool(self.bot_token and self.chat_id)
+
+    def send_message(self, text: str) -> bool:
+        """Send a message to the configured chat. Returns True on success."""
+        if not self.is_configured():
+            logger.warning("Telegram not configured (missing BOT_TOKEN or CHAT_ID)")
+            return False
+
+        if len(text) > MAX_TEXT_LENGTH:
+            text = text[: MAX_TEXT_LENGTH - 20] + "\n\n[...truncated]"
+
+        try:
+            import httpx
+            response = httpx.post(
+                TELEGRAM_API_URL.format(token=self.bot_token),
+                json={
+                    "chat_id": self.chat_id,
+                    "text": text,
+                    "parse_mode": "HTML",
+                },
+                timeout=15.0,
+            )
+            if response.status_code == 200 and response.json().get("ok"):
+                logger.info("Telegram message sent successfully")
+                return True
+            else:
+                logger.error(f"Telegram API error: {response.text[:200]}")
+                return False
+        except Exception as e:
+            logger.error(f"Telegram send failed: {e}")
+            return False

pyproject.toml

@@ -17,6 +17,7 @@ dependencies = [
     "jinja2>=3.1",
     "pyyaml>=6.0",
     "python-dotenv>=1.0",
+    "playwright-stealth>=2.0.2",
 ]
 
 [project.scripts]

sites/aaoinfo.yaml

@@ -0,0 +1,60 @@
+site_id: aaoinfo
+enabled: true
+repair_needed: false
+
+identity:
+  label: AAO Career Center
+  category: aggregator
+  base_url: https://careers.aaoinfo.org
+  start_url_template: "https://careers.aaoinfo.org/jobs/"
+
+browser:
+  profile_mode: persistent_chrome_profile
+  profile_name: JobAgent
+  headed_on_learn: true
+  headless_on_run: true
+
+search:
+  keyword_mode: none
+  location_mode: none
+  sort_mode: none
+  date_mode: none
+  result_list_wait_selector: "div.job-tile"
+
+pagination:
+  mode: none
+  max_pages: 10
+
+extract:
+  container_selectors:
+    - "div.job-tile:not(.job-mo)"
+  fields:
+    title:
+      text:
+        - ".job-title a"
+    company:
+      text:
+        - ".job-company-row"
+    location:
+      text:
+        - ".job-location"
+    posted_text:
+      text:
+        - ".job-posted-date"
+    url:
+      attr:
+        selector: ".job-title a"
+        name: href
+
+post_filters:
+  include_posted_text: []
+
+healthcheck:
+  min_items_expected: 0
+  required_fields:
+    - title
+  fail_if_zero_when_known_active: false
+
+learn:
+  last_learned_at: null
+  source_url: null

sites/aroragroup.yaml

@@ -0,0 +1,56 @@
+site_id: aroragroup
+enabled: true
+repair_needed: false
+
+identity:
+  label: Arora Group Jobs
+  category: employer
+  base_url: https://jobs.aroragroup.com
+  start_url_template: "https://jobs.aroragroup.com/"
+
+browser:
+  profile_mode: persistent_chrome_profile
+  profile_name: JobAgent
+  headed_on_learn: true
+  headless_on_run: true
+
+search:
+  keyword_mode: none
+  location_mode: none
+  sort_mode: none
+  date_mode: none
+  result_list_wait_selector: ".job-post-row"
+
+pagination:
+  mode: none
+  max_pages: 1
+
+extract:
+  container_selectors:
+    - ".job-post-row[onclick]"
+  fields:
+    title:
+      text:
+        - ".POST_TITLE"
+        - "h3.job-post-title a"
+    location:
+      text:
+        - ".POST_LOCATION"
+        - ".job-post-location"
+    url:
+      attr:
+        selector: "a.job-post-href"
+        name: href
+
+post_filters:
+  include_posted_text: []
+
+healthcheck:
+  min_items_expected: 0
+  required_fields:
+    - title
+  fail_if_zero_when_known_active: false
+
+learn:
+  last_learned_at: null
+  source_url: null

sites/docshealth.yaml

@@ -0,0 +1,64 @@
+site_id: docshealth
+enabled: true
+repair_needed: false
+
+identity:
+  label: DOCS Health
+  category: employer
+  base_url: https://recruiting.paylocity.com
+  start_url_template: "https://recruiting.paylocity.com/recruiting/jobs/All/4f5b87cb-fda9-4243-976b-499bbf0ce9fc/DOCS-Health?search={keywords_urlencoded}&location=All%20Locations&department=All%20Departments"
+
+browser:
+  profile_mode: persistent_chrome_profile
+  profile_name: JobAgent
+  headed_on_learn: true
+  headless_on_run: true
+
+search:
+  keyword_mode: url_param
+  location_mode: url_param
+  sort_mode: none
+  date_mode: none
+  result_list_wait_selector: ".job-listing-job-item"
+
+search_override:
+  keywords:
+    - dentist
+    - orthodontics
+    - orthodontist
+  multi_keyword_mode: separate
+
+pagination:
+  mode: none
+  max_pages: 1
+
+extract:
+  container_selectors:
+    - "div.job-listing-job-item"
+  fields:
+    title:
+      text:
+        - ".job-item-title a"
+    location:
+      text:
+        - ".location-column .job-item-normal"
+    posted_text:
+      text:
+        - ".job-title-column > span:not(.job-item-title):not(.job-item-normal)"
+    url:
+      attr:
+        selector: ".job-item-title a"
+        name: href
+
+post_filters:
+  include_posted_text: []
+
+healthcheck:
+  min_items_expected: 0
+  required_fields:
+    - title
+  fail_if_zero_when_known_active: false
+
+learn:
+  last_learned_at: null
+  source_url: null

sites/global.yaml

@@ -5,7 +5,7 @@ runtime:
   slow_mo_ms: 0
   default_timeout_ms: 15000
   navigation_timeout_ms: 30000
-  max_pages_per_site: 3
+  max_pages_per_site: 10
   min_delay_ms: 1200
   max_delay_ms: 3500
 

sites/hospitaljobsonline.yaml

@@ -0,0 +1,60 @@
+site_id: hospitaljobsonline
+enabled: true
+repair_needed: false
+
+identity:
+  label: Hospital Jobs Online
+  category: aggregator
+  base_url: https://www.hospitaljobsonline.com
+  start_url_template: "https://www.hospitaljobsonline.com/jobs/q-korea-jobs//?page=1&limit=100"
+
+browser:
+  profile_mode: persistent_chrome_profile
+  profile_name: JobAgent
+  headed_on_learn: true
+  headless_on_run: true
+
+search:
+  keyword_mode: none
+  location_mode: none
+  sort_mode: none
+  date_mode: none
+  result_list_wait_selector: "div.jobresult"
+
+pagination:
+  mode: url_increment
+  max_pages: 10
+
+extract:
+  container_selectors:
+    - "div.jobresult"
+  fields:
+    title:
+      text:
+        - ".jobtitle a"
+    company:
+      text:
+        - ".location a"
+    location:
+      text:
+        - ".location"
+    posted_text:
+      text:
+        - ".location .date"
+    url:
+      attr:
+        selector: ".jobtitle a"
+        name: href
+
+post_filters:
+  include_posted_text: []
+
+healthcheck:
+  min_items_expected: 0
+  required_fields:
+    - title
+  fail_if_zero_when_known_active: false
+
+learn:
+  last_learned_at: null
+  source_url: null

sites/indeed.yaml

@@ -1,5 +1,5 @@
 site_id: indeed
-enabled: true
+enabled: false
 repair_needed: false
 
 identity:
@@ -12,7 +12,7 @@ browser:
   profile_mode: persistent_chrome_profile
   profile_name: JobAgent
   headed_on_learn: true
-  headless_on_run: true
+  headless_on_run: false
 
 search:
   keyword_mode: url_param

sites/linkedin.yaml

@@ -0,0 +1,74 @@
+site_id: linkedin
+enabled: true
+repair_needed: false
+
+identity:
+  label: LinkedIn Jobs
+  category: aggregator
+  base_url: https://www.linkedin.com
+  start_url_template: "https://www.linkedin.com/jobs/search/?keywords={keywords_urlencoded}&f_TPR=r86400&sortBy=DD"
+
+browser:
+  profile_mode: persistent_chrome_profile
+  profile_name: JobAgent
+  headed_on_learn: true
+  headless_on_run: false
+
+search:
+  keyword_mode: url_param
+  location_mode: url_param
+  sort_mode: url_param
+  date_mode: url_param
+  result_list_wait_selector: ".jobs-search__results-list li, .scaffold-layout__list-container li"
+
+pagination:
+  mode: next_button
+  next_button_selectors:
+    - "button[aria-label='View next page']"
+    - "button[aria-label*='next']"
+    - "li.artdeco-pagination__indicator--number.selected + li button"
+  max_pages: 3
+
+extract:
+  container_selectors:
+    - ".jobs-search__results-list li"
+    - ".scaffold-layout__list-container li"
+    - "li:has(.base-card)"
+  fields:
+    title:
+      text:
+        - "h3.base-search-card__title"
+        - ".job-card-list__title"
+        - "h3"
+        - "a[data-control-name='job_card_title']"
+    company:
+      text:
+        - "h4.base-search-card__subtitle"
+        - ".job-card-container__primary-description"
+        - "a.job-card-container__company-name"
+    location:
+      text:
+        - "span.job-search-card__location"
+        - ".job-card-container__metadata-item"
+    posted_text:
+      text:
+        - "time"
+        - ".job-search-card__listdate"
+        - "span[class*='listdate']"
+    url:
+      attr:
+        selector: "a.base-card__full-link, a[data-control-name='job_card_title'], a[href*='/jobs/view/']"
+        name: href
+
+post_filters:
+  include_posted_text: []
+
+healthcheck:
+  min_items_expected: 0
+  required_fields:
+    - title
+  fail_if_zero_when_known_active: false
+
+learn:
+  last_learned_at: null
+  source_url: null

sites/pdshealth.yaml

@@ -0,0 +1,57 @@
+site_id: pdshealth
+enabled: true
+repair_needed: false
+
+identity:
+  label: PDS Health
+  category: employer
+  base_url: https://jobs.pdshealth.com
+  start_url_template: "https://jobs.pdshealth.com/jobs?keywords=orthodontist&sortBy=relevance&page=1&limit=100"
+
+browser:
+  profile_mode: persistent_chrome_profile
+  profile_name: JobAgent
+  headed_on_learn: true
+  headless_on_run: true
+
+search:
+  keyword_mode: none
+  location_mode: none
+  sort_mode: none
+  date_mode: none
+  result_list_wait_selector: "mat-expansion-panel.search-result-item"
+
+pagination:
+  mode: url_increment
+  max_pages: 20
+
+extract:
+  container_selectors:
+    - "mat-expansion-panel.search-result-item"
+  fields:
+    title:
+      text:
+        - "a.job-title-link"
+    company:
+      text:
+        - ".label-value.tags1"
+    location:
+      text:
+        - ".label-value.location"
+    url:
+      attr:
+        selector: "a.job-title-link"
+        name: href
+
+post_filters:
+  include_posted_text: []
+
+healthcheck:
+  min_items_expected: 0
+  required_fields:
+    - title
+  fail_if_zero_when_known_active: false
+
+learn:
+  last_learned_at: null
+  source_url: null

sites/saltdental.yaml

@@ -0,0 +1,57 @@
+site_id: saltdental
+enabled: true
+repair_needed: false
+
+identity:
+  label: Salt Dental Partners
+  category: employer
+  base_url: https://jobs.saltdentalpartners.com
+  start_url_template: "https://jobs.saltdentalpartners.com/jobs?keywords=orthodontist&sortBy=relevance&page=1"
+
+browser:
+  profile_mode: persistent_chrome_profile
+  profile_name: JobAgent
+  headed_on_learn: true
+  headless_on_run: true
+
+search:
+  keyword_mode: none
+  location_mode: none
+  sort_mode: none
+  date_mode: none
+  result_list_wait_selector: "mat-expansion-panel.search-result-item"
+
+pagination:
+  mode: url_increment
+  max_pages: 5
+
+extract:
+  container_selectors:
+    - "mat-expansion-panel.search-result-item"
+  fields:
+    title:
+      text:
+        - "a.job-title-link"
+    company:
+      text:
+        - ".label-value.tags1"
+    location:
+      text:
+        - ".label-value.location"
+    url:
+      attr:
+        selector: "a.job-title-link"
+        name: href
+
+post_filters:
+  include_posted_text: []
+
+healthcheck:
+  min_items_expected: 0
+  required_fields:
+    - title
+  fail_if_zero_when_known_active: false
+
+learn:
+  last_learned_at: null
+  source_url: null

sites/southernortho.yaml

@@ -0,0 +1,57 @@
+site_id: southernortho
+enabled: true
+repair_needed: false
+
+identity:
+  label: Southern Orthodontic Partners
+  category: employer
+  base_url: https://recruiting.paylocity.com
+  start_url_template: "https://recruiting.paylocity.com/recruiting/jobs/All/06144481-7908-457c-9312-51cf7b58cf01/Southern-Orthodontic-Partners-Mgmt-LLC?search=orthodontist&location=All%20Locations&department=All%20Departments"
+
+browser:
+  profile_mode: persistent_chrome_profile
+  profile_name: JobAgent
+  headed_on_learn: true
+  headless_on_run: true
+
+search:
+  keyword_mode: none
+  location_mode: none
+  sort_mode: none
+  date_mode: none
+  result_list_wait_selector: ".job-listing-job-item"
+
+pagination:
+  mode: none
+  max_pages: 1
+
+extract:
+  container_selectors:
+    - "div.job-listing-job-item"
+  fields:
+    title:
+      text:
+        - ".job-item-title a"
+    company:
+      text:
+        - ".location-column .job-item-normal"
+    posted_text:
+      text:
+        - ".job-title-column > span:not(.job-item-title):not(.job-item-normal)"
+    url:
+      attr:
+        selector: ".job-item-title a"
+        name: href
+
+post_filters:
+  include_posted_text: []
+
+healthcheck:
+  min_items_expected: 0
+  required_fields:
+    - title
+  fail_if_zero_when_known_active: false
+
+learn:
+  last_learned_at: null
+  source_url: null

sites/usajobs.yaml

@@ -0,0 +1,70 @@
+site_id: usajobs
+enabled: true
+repair_needed: false
+
+identity:
+  label: USAJOBS
+  category: government
+  base_url: https://www.usajobs.gov
+  start_url_template: "https://www.usajobs.gov/Search/Results?k={keywords_urlencoded}&p=1"
+
+browser:
+  profile_mode: persistent_chrome_profile
+  profile_name: JobAgent
+  headed_on_learn: true
+  headless_on_run: true
+
+search:
+  keyword_mode: url_param
+  location_mode: url_param
+  sort_mode: url_param
+  date_mode: url_param
+  result_list_wait_selector: ""
+
+pagination:
+  mode: next_button
+  next_button_selectors:
+    - "button#page-m-next"
+    - "button[title='Go to next page']"
+  max_pages: 10
+
+extract:
+  container_selectors:
+    - "#search-results .border.border-gray-lighter"
+  fields:
+    title:
+      text:
+        - "h2 a"
+    company:
+      text:
+        - "p strong"
+    location:
+      text:
+        - "div:nth-child(2) div:first-child p:nth-child(3)"
+    posted_text:
+      text:
+        - ".italic"
+    url:
+      attr:
+        selector: "h2 a"
+        name: href
+
+search_override:
+  keywords:
+    - dentist
+    - orthodontics
+    - orthodontist
+  multi_keyword_mode: separate
+
+post_filters:
+  include_posted_text: []
+
+healthcheck:
+  min_items_expected: 0
+  required_fields:
+    - title
+  fail_if_zero_when_known_active: false
+
+learn:
+  last_learned_at: null
+  source_url: null

uv.lock

@@ -156,6 +156,7 @@ dependencies = [
     { name = "jinja2" },
     { name = "loguru" },
     { name = "playwright" },
+    { name = "playwright-stealth" },
     { name = "pydantic" },
     { name = "pydantic-settings" },
     { name = "python-dotenv" },
@@ -179,6 +180,7 @@ requires-dist = [
     { name = "jinja2", specifier = ">=3.1" },
     { name = "loguru", specifier = ">=0.7" },
     { name = "playwright", specifier = ">=1.40" },
+    { name = "playwright-stealth", specifier = ">=2.0.2" },
     { name = "pydantic", specifier = ">=2.0" },
     { name = "pydantic-settings", specifier = ">=2.0" },
     { name = "python-dotenv", specifier = ">=1.0" },
@@ -431,6 +433,18 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/c8/c4/cc0229fea55c87d6c9c67fe44a21e2cd28d1d558a5478ed4d617e9fb0c93/playwright-1.58.0-py3-none-win_arm64.whl", hash = "sha256:32ffe5c303901a13a0ecab91d1c3f74baf73b84f4bedbb6b935f5bc11cc98e1b", size = 33085919, upload-time = "2026-01-30T15:09:45.71Z" },
 ]
 
+[[package]]
+name = "playwright-stealth"
+version = "2.0.2"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "playwright" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/61/ee/871901103c7b2a12070011fd4d978191f8f962837bf8bb51847274f528fa/playwright_stealth-2.0.2.tar.gz", hash = "sha256:ac57e51873190da5e653e03720e948c8f0a3d06b098f1d56763103d23ee48143", size = 24902, upload-time = "2026-02-13T02:36:25.137Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/f1/30/f95f087f4b071611a7f63a2a0c9af4df3ac046dae2a693bfdacd70512867/playwright_stealth-2.0.2-py3-none-any.whl", hash = "sha256:37a5733f481b9c0ad602cf71491aa5a7c96c2a2fe4fa1e7ab764d2cd35520f2f", size = 33209, upload-time = "2026-02-13T02:36:26.334Z" },
+]
+
 [[package]]
 name = "pluggy"
 version = "1.6.0"