Loading…
Dwg 02 · Config Sheet

Edit Preferences

These settings control what the discovery engine searches for and filters out on every run. Everything here stays editable.

Apify Connection (for real LinkedIn searches)
Stays in this browser tab only — used to call Apify's API directly from your device. It's never sent to Claude and isn't saved anywhere once you close the tab. Get a token from your Apify console → Integrations. The account also needs enough usage headroom — this actor charges per job scraped.
Bright Data LinkedIn Jobs Scraper (fallback if Apify isn't set up)
Uses Bright Data's LinkedIn Jobs Scraper — a real scraper that reads the actual LinkedIn job page (not a search snippet), so posted-date and applicant-count come back reliably. Sign up free at brightdata.com (5,000 free records/month, no card needed), then find your Dataset ID for the LinkedIn Jobs Scraper product on your dashboard.
How it works: triggers a scraping job, then checks back every 4 seconds until it's done — real runs have taken anywhere from under a minute up to 6+ minutes, so it waits up to 20 minutes total before giving up. You'll see a live "Xm Ys elapsed" status while it waits, so you can tell it's still working rather than stuck. This uses several short requests instead of one long one, since a single long request would time out on most shared hosting.
Balance protection: checks your real Bright Data account balance before starting, and again before each saved keyword. Under $0.50, you'll get a warning but the search still runs. At $0.00, it stops immediately and won't spend anything further — a red popup tells you exactly what happened. Save & Test also shows your current balance.
What actually gets searched: Job Keywords and Location drive the search directly. Time Posted and Internship type are sent to Bright Data as real search filters (not just applied afterward). Hard Exclusion Rules are sent too, as a native exclude-list Bright Data filters at the source — on top of still being hidden here if one slips through. Re-running a search on a job you already have refreshes its posted-time and applicant-count instead of ignoring it.
(This replaces the Google/Serper snippet-based fallback — Bright Data scrapes the real page instead of reading search results.)
RocketReach Connection (for recruiter email lookups)
Also stays in this browser tab only — used to call RocketReach's API directly from your device, never sent to Claude. Get a key from RocketReach → Account → API Usage & Settings. Each lookup consumes a credit on your RocketReach plan, so searches in Board 2 are limited by your account's usage, not by this site.
Anthropic API Connection (for Tailor Resume, when hosted outside Claude)
Left visible (not masked) so you can actually see what got pasted — a stray space or line break from copy-paste is a common reason this silently fails. Also stays in this browser tab only. Get a key from the Anthropic Console; it bills to whatever account that key belongs to (usage-based, typically a fraction of a cent per resume).
Why this exists: when this page is open inside Claude, Tailor Resume can call the model without any key. Once the site is hosted on its own domain, that free path isn't available, so a real key here is what makes the feature keep working after hosting.
Resume
📎 Upload Resume / CV (.txt or .docx parse automatically — used for matching, tailoring & ATS optimization)
No resume text loaded yet — tailoring will stay a placeholder until one is added.
Location Preference
Selection rule: choose one or both. — set by you.
Work Arrangement
Multiple or all allowed. Sent to Apify's own search when Apify is the source. Not filtered for Bright Data results — job titles almost never say "Remote"/"Hybrid"/"Onsite" explicitly, so guessing here risks hiding jobs that actually match.
Term Preference
Real filter now: hides a job only if its title clearly names a different season than any you've checked (e.g. title says "Summer 2027" but Summer is unchecked). Jobs whose title doesn't mention a season at all are never hidden by this.
Target Year(s)
Same safe rule: only hides a job if its title clearly names a year you haven't checked. No year mentioned in the title = never hidden.
Opportunity Type
Strict filter: a job is only shown if its title actually says "Intern"/"Internship" or "Co-op" — whichever you've checked. Unlike the other filters below, a title that doesn't mention either gets hidden too, not just a clear conflict. Also sent to Apify's own search when Apify is the source.
Work Schedule
Same safe rule: hides a job only if its title explicitly says "Part-Time" or "Full-Time" and that one isn't checked.
Compensation Preference
Saved, but honestly not filtered — whether a role is paid almost never appears in the job title, and guessing from the full description risks hiding jobs that are actually paid. Worth checking manually per posting.
Time Posted Preference
Real filter now, in both directions: sent to Bright Data/Apify as the actual search window, AND applied live to Board 1 — a job already on your board gets hidden the moment it's older than whichever option is selected here, recalculated continuously as real time passes. Single choice only (radio, not checkboxes) — the previous multi-checkbox version silently only ever used one of them anyway.
Competition Filter
Set by you: jobs 7+ days old with more than 50 applicants are hidden from Board 1. Jobs under that applicant count still show even if older. Edit the numbers above anytime.
Button Cooldown
Stops accidental double-clicks and spam-clicking from firing a button's action multiple times in a row — mainly matters for buttons that cost money or credits (Run Now, Save & Test, RocketReach Search). Doesn't apply to simple UI toggles like show/hide password or window controls.
Job Keywords
One search is run per keyword. Comma-separated.
With a long keyword list, doing one at a time means a lot of manual clicking. Raise this to process several keywords back-to-back in a single click — it still stops completely and waits for you afterward, it just does more before stopping. Type a number higher than your keyword count (e.g. 999) to run the whole list in one go.
Additional Notes
Still feeds Tailor Resume generation as before. Now also actually read for two things: (1) "avoid" signals — extracted as real exclusion terms and enforced like Hard Exclusion Rules below (hidden here AND sent to Bright Data's search), and (2) an explicit request to match jobs by your resume rather than only literal keywords (e.g. "use my resume to find relevant jobs even if the exact title isn't there") — when detected, each new search result is checked against your saved resume text, keeping jobs that are a genuine fit even if the title doesn't match your keywords. Both require an Anthropic API key (set further down) and, for the resume-matching part, resume text saved under Tailor Resume settings.
Hard Exclusion Rules (never show)
Real filter: comma-separated terms. Any job whose title or company contains one of these is hidden from Board 1 — check the legend line above the table for how many are currently excluded.
Manual Search Only
The automatic scheduling system has been removed — this board now only searches when you click Run Now (or Run Again after a completed search). Nothing runs on its own, so it never spends Apify or Bright Data credits without you asking.
Import Applied Jobs from LinkedIn (Bookmarklet)
What this is: a one-click bookmark that reads whatever is already on your screen when you're manually logged into LinkedIn and viewing Jobs → My Jobs → Applied (or the newer "Job Tracker"). It does not log in for you, does not store your LinkedIn password, and doesn't automate anything about visiting the page — you do that part yourself, like any normal browsing.

Be honest about the limits: LinkedIn has no public API for this data, and their page layout can change anytime without notice — this scrapes based on durable patterns (actual job links), not exact visual structure, but it can still break. If it stops finding jobs, that likely means LinkedIn changed something and this needs updating.
A separate credential from your main login — lives inside the bookmarklet code sitting in your browser's bookmarks bar, so it's kept apart from your actual site password on purpose. Regenerate it anytime if you're ever unsure it's still private.
After clicking the bookmarklet on LinkedIn, come back here and paste (Ctrl/Cmd+V) into the box below, then click Import.
Maintenance Mode
— checking —
Locks the whole site for every visitor — including you, on any device — behind a separate developer-only screen. While it's on, the regular ID/password login won't work at all; only the Developer ID/Password on that screen gets in. Turn it back off from here (or from the developer screen once you're in).

Update Summary

Result of this discovery-engine cycle.

Semi Auto Fill

Repetitive fields are filled for you and your approved resume is attached. The system stops here — you review and submit on the job site yourself.

No tailored resume approved yet for this job — using master resume placeholder
Referral requests, recruiter outreach, and final submission are never automated — they need your sign-off.
Recruiter Lookup

RocketReach Search

Search for this recruiter's email.

Sheet 03 · Notes

Notes

Full note for this application. Nothing here is shared anywhere.

Tailor Resume