Grow your professional network and run your job search from one calm, honest workspace.
Know who to contact next. Find real, recent jobs from employers' own career pages. Track every application, and every follow-up.
A 50-second tour on fictional example data (720p preview). Watch it in 1080p (MP4). Regenerate it with npm run docs:video.
You stay in control
The app never scrapes LinkedIn, never sends invitations or messages for you and never asks for your LinkedIn password. You open every profile and send every message yourself. It only reads what you paste, import or search for in public sources, and it tells you where every fact came from.
Find your way around
Quick links: Open the live app · Watch the demo · Detailed feature guide · Deployment guide · Security reporting
Showcase
Four things that make this different from a spreadsheet.
1. Quick capture: paste a LinkedIn search page, get a clean list
On a LinkedIn people search, press Ctrl+A, Ctrl+C. Back in the app, press Ctrl+V. Everyone on the page appears with their school and program already matched to your own lists, ready to add in one click. Nothing is fetched from LinkedIn: only what you copied is read.
- Matches schools and programs from the wording on the page, using the aliases you set (so "UBC", "U of T" and "UofT" mean what you say they mean).
- Skips people you already have and people you're already connected to; flags rows that need a school or program instead of guessing.
- After you add the page, the cursor goes back to the paste box: the next page is one Ctrl+V away.
2. Discover: people worth meeting, with the receipts
Ask in plain language: "Find 10 UBC alumni who work at Amazon." A research assistant searches public sources (GitHub, ORCID and optional web search), then verifies the best matches on their own pages. It never visits LinkedIn, and it works even when you name no school ("software engineers at startups in Vancouver").
- Every fact quotes its source. The server checks each quote against the page and drops anything it can't find.
- Scores are rules, not vibes. Education, program, career, company, location and evidence each add a visible number you can inspect.
- It returns fewer people rather than lowering the bar. Lookalike schools ("New York University" is not "York University"), people with no sign of the company you named, people outside a country you asked for and people you already know are set aside, with the reason shown.
- It says how much the public sources know. "ORCID lists 10 people matching Simon Fraser University + Microsoft" explains a short list. A job on a profile nobody has edited for two years is only a possible match, and a city taken from an employer is labelled "Employer's office".
- Free to run. AI goes through a chain of free services (two Gemini models, then Groq and Mistral if you add their free keys) and falls back to plain evidence rules when they are used up.
3. Jobs: real postings, verified links, fit explained
Search recent jobs in any country and any field. Postings come from employers' own job boards (Greenhouse, Lever, Ashby, SmartRecruiters, Workable, Recruitee and Workday) plus free remote-job feeds, and every link is checked before Apply turns on.
- Verified, or it doesn't get an Apply button. Links come from the source's own data, never from a pattern or a model.
- Fit is explained. Each requirement is matched against your resume and the card says why: strong matches, what's missing, what's preferred.
- Tailor your resume honestly. AI rewrites a copy only if an accuracy check can back every line; otherwise you get specific improvement notes. Nothing is faked.
- Save a search to get alerts in the app (and by email, if you turn that on).
- Compare, hide, note. Tick two or three jobs to compare them side by side, hide a company or a title word (with Undo), and keep a private note on each job.
- A resume per kind of role. Keep separate base resumes (Software, Data, Product…); each job shows which one fits best and why.
4. Honest progress: pending never counts
Set your own goal, schools, disciplines and targets. The Overview shows where you stand against each one. Only accepted invitations count toward a target, and pending ones are shown apart in a hatched segment, so the numbers never flatter you.
以上为 README 的开头部分。其余内容请阅读原文。 查看完整 README ↗
