Screen your whole pipeline in a fraction of the clicks — AI candidate scoring, one-click tagging, and bulk rejection emails, right on your Greenhouse candidate-review pages.
What it is
You know the drill: a dozen tabs open, a résumé buried three clicks deep, and a pile of “I’ll get back to them” candidates you never quite do. GreenMaxing Toolbar is a side panel that docks itself the moment you open a candidate-review page in Greenhouse and puts every move you actually reach for in one place. Instead of clicking through tabs and buried menus one candidate at a time, you move through a whole stage in a fraction of the clicks — sharper decisions, made faster, and a real answer back to everyone who applied.
Candidate snapshot — everything you’d normally squint across the page for, up front: name, source, salary, work authorization, location, and tags. Open the résumé in one click (no download) and dock the candidate’s LinkedIn right beside your work — no tab-hunting mid-review.
AI candidate scoring — a 0–100 fit score with a plain-English rationale and a per-criterion breakdown, judged against the criteria you set (job description, need-to-haves, hard dealbreakers, soft negatives). Runs on your own AI key (Claude, GPT, or Gemini), opt-in and consent-gated, with optional auto-scoring as you view.
Auto-rank a stage — score and rank an entire stage in one pass. See the time and rough cost before it runs, calibrate on two candidates so the results match your judgment, then let it auto-tag the top matches above a threshold you choose. A stack of applicants becomes a ranked shortlist while you grab coffee.
Quick-tags & Mark Reviewed — sort candidates as fast as you can read them with one-click tags you configure. Applied tags stay highlighted so you always know where you left off, and the panel can auto-advance to the next candidate the moment you tag.
Reject Tagged — close the loop with everyone — the one that changes your week. Batch-reject candidates grouped by the quick-tags you applied, and give each tag-group its own rejection email: type it, load one of your Greenhouse templates, or draft and refine it with AI. Toggle “reject” and “send email” independently per group — keep a “Potential” group untouched, or reject a group without emailing anyone. Review every message, then send them all in one pass through Greenhouse, behind a count-confirmation and an enable-once safety gate. Real, personal closure for the whole list — in the time it used to take to write one.
Email templates — stop rewriting the same message. Save or update your Greenhouse email templates straight from the panel, with an AI assistant on hand to write a new one or sharpen what you’ve got.
It runs entirely inside your own logged-in browser session, on your own AI key — no separate account, no company server behind it (see Your data & privacy). Every candidate gave you their time; GreenMaxing helps you give them a real answer back — faster, without cutting the human part.
Install & find it
Install GreenMaxing Toolbar from the Chrome Web Store and, if you like, pin it from Chrome's puzzle-piece menu.
Make sure you're signed in to Greenhouse in the same Chrome profile (and to LinkedIn too, if you want the LinkedIn popup).
Open any Greenhouse candidate-review page — the panel docks itself to the side of the screen automatically. There's nothing to click to launch it.
Where it shows up
The panel appears on Greenhouse review pages (the screen where you page through candidates in a stage). If you don't see it, refresh the Greenhouse tab.
First-time setup
Everything is configured from the panel's settings. Click the ⚙ gear at the top of the panel to open it. There are three things to set up — the first one unlocks AI scoring.
Step 1 — Turn on AI scoring (provider, key, consent)
In ⚙ settings, open the AI scoring section and pick your provider: Anthropic (Claude), OpenAI (GPT), or Google (Gemini).
Paste that provider's API key — the one from your own account with that provider. It's stored only on your computer.
Tick the consent checkbox. This confirms you understand that, when you score a candidate, their résumé text and name are sent to the AI provider you picked, under your own key and that provider's terms. Scoring stays switched off until both a key and consent are in place.
You bring your own key
GreenMaxing has no AI service of its own. You use your own account with Anthropic, OpenAI, or Google, and the AI usage is billed to you. Your key never leaves your computer for anything except talking to that provider.
Step 2 — Add your quick-tags
Still in ⚙ settings, add the quick-tags you use to sort candidates — for example Reject – Experience or Reject – Location. These do two jobs:
They become the one-click tag buttons in the panel.
They're what Reject Tagged uses to find the candidates to reject in a batch.
Step 3 — (Optional) Fill in per-job scoring criteria
To make AI scores sharper for a specific role, add scoring criteria for that job. The job description auto-pulls from the live Greenhouse job post (or you can paste your own), and you add:
Need-to-haves — the core things the candidate should have.
Hard dealbreakers — objective auto-rejects that cap the score very low.
Soft negatives / notes — preferences that nudge the score down but never sink it on their own.
You can skip this and still get a general score — the criteria just make it more tailored to the role.
Everyday use
Score a candidate
Open a candidate on a review page.
Click Score candidate in the panel. GreenMaxing reads the résumé, combines it with the job description and your criteria, and asks your AI provider for a 0–100 match score with a short rationale.
The score card appears in the panel. Optionally, the result can be written to a private, admin-only Greenhouse note and a tag can be auto-applied when the score is high.
Auto-rank a whole stage
Open the stage you want to rank.
Choose Auto-rank. GreenMaxing scores the loaded candidates so you can review the strongest first.
Because this sends several résumés to your AI provider at once, you'll get a quick note telling you how many résumés will be sent before it runs.
Save or AI-edit an email template
Open the template tools in the panel and write or paste your email.
Save a new template, or Update an existing one.
Use the AI write/edit assistant to draft or refine the wording — for example "make this friendlier" or "shorten it." You review the result before anything is used; nothing is sent to a candidate automatically.
Reject Tagged (sends real rejection emails)
Reject Tagged finds every candidate in the stage that carries one of your reject quick-tags, rejects them in Greenhouse, and sends each one their Greenhouse rejection email.
⚠ This sends real emails and can't be undone
Reject Tagged sends actual rejection emails to real candidates through Greenhouse — the same as rejecting them by hand. It is irreversible. Because of that:
The very first time, you must complete a one-time "go live" opt-in confirming you understand it sends real emails. Until you do, it stays in a safe, no-send state.
Every run then asks you to confirm before anything is sent.
Always double-check which candidates carry your reject tags first. If a candidate's email or eligibility can't be confirmed, GreenMaxing skips them rather than guessing.
Your data & privacy
GreenMaxing has no server of its own — no backend, no analytics, no developer database. Everything runs in your own browser and your own Greenhouse session. Your settings, your API key, your criteria, and any AI scores are stored locally on your device only.
The only data that leaves your machine is when you use an AI feature: the candidate's résumé text and name plus your role criteria are sent directly to the AI provider you chose, under your own key and their terms. Screening-question answers are stripped out before that request is made, and the developer never receives any of it.
You can clear locally stored scores and data at any time from settings, and removing the extension deletes its local storage.
Full details are in the Privacy Policy — please read it before turning on AI features.
Support
Questions, bug reports, or feedback? We're happy to help.
When reporting a problem, it helps to include which browser you're using and a rough description of what you were doing when it happened. Please don't include real candidate personal data in a support email.