An open-source, AI-native ATS that runs entirely inside your own Google account — free forever.
No servers. No database bills. No per-seat pricing. You paste six files into a free Google Apps Script project, run one function, and you have a working applicant tracking system: a hiring web app for your team, a public careers page for candidates, AI resume parsing and scoring, interview scheduling, debriefs, and analytics — all stored in a Google Sheet and Drive folder that you own.
Recruiting software is overpriced and locked-in. Small teams pay hundreds of dollars per month for tools whose core job is: keep a list of candidates, read their CVs, schedule interviews, and collect feedback. Your data lives on someone else's servers, exports are painful, and the AI features cost extra.
AgentATS is a gift to the HR and talent-acquisition community. It exists to prove that a genuinely useful, AI-assisted ATS can be free, transparent, and owned end-to-end by the people using it. Read every line of code. Change anything. Your candidate data never leaves your Google account.
Requisitions & pipeline
- Full requisition management: title, department, location, level, openings, salary band, priority, status, JD attachment, hiring manager and recruiter ownership.
- Kanban-style pipeline with configurable stages, stage-change tracking, and "stuck candidate" SLA alerts on your Today view.
- Archiving keeps large pipelines fast while preserving a year of live data for annual metrics, with search and restore.
Candidate intake
- Public careers page (
?page=apply) — candidates apply with a CV upload; submissions are capped and size-limited to resist abuse. - CV-by-email: point a
careers@mailbox at the included forwarder script and every emailed CV is parsed and added automatically — with retry logic so no CV is ever silently lost. - Manual add, bulk upload, and chat-style commands ("Add Asha Rao, name@example.com, backend engineer").
AI resume parsing & scoring
- Gemini-powered CV parsing: name, contact, experience, skills, notice period, compensation, work authorization and more, straight into structured columns.
- A 37-category scoring rubric (
TalentRubric.gs) with role archetypes and company-type modifiers — tune weights in plain English or per category, with a full change log of who changed what. - Match grades and fit scores per requisition, benchmark generation from your own hiring bar, and stack ranking.
- Bias-masked screening: evaluate candidates with identifying details hidden.
Interviews & decisions
- Interview scheduling with Google Calendar + Meet links, RSVP tracking, and per-requisition interview plans (each role gets its own rounds).
- Structured feedback via Google Forms or Slack-style slash commands.
- Debrief view and a candidate packet PDF for hiring committees.
Sourcing & analytics
- Sourcing engine: live opt-in sourcing (e.g. Hacker News "Who wants to be hired") plus an X-ray query builder for LinkedIn, GitHub and resume search.
- Recruiting analytics per hiring manager and per role, interview round metrics, and a one-click Looker Studio data tab for live dashboards.
Team, trust & safety
- Role-based access: Admin, Recruiter, Hiring Manager, Interviewer — enforced server-side, with personal access links per teammate.
- Audit trail of changes; formula-injection and abuse protections; secrets kept in Script Properties, never in code.
- Notifications by email or Google Chat webhook; automated backups of your tracker.
AgentATS uses Google Gemini through your own API key — bring your own key, pay as you go. Parsing and scoring calls cost cents, not seats: a busy month of screening typically costs less than a coffee. The key lives in a Script Property (GEMINI_KEY), is never written into code, and never touches any third-party server — calls go directly from your Google account to Google's API. The prompt layer is model-agnostic by design; swapping in another provider is a small, well-contained change.
Honest benchmarking against the tools this replaces for small and mid-size teams:
| AgentATS | Leading paid ATS platforms | |
|---|---|---|
| Price | Free, forever (plus cents of AI usage on your own key) | Roughly $300–$1,000+/month for small teams; enterprise pricing beyond |
| Data ownership | 100% in your Google account — the "database" is a Sheet you can open | Vendor cloud; exports on request |
| AI screening & scoring | Included, transparent 37-category rubric you can read and tune | Paid add-ons or opaque scoring |
| Careers page | Included (?page=apply) |
Included |
| Email CV intake | Included (forwarder script) | Included |
| Scheduling | Google Calendar + Meet, RSVP tracking | Deeper multi-calendar orchestration |
| Customization | Edit the source | Config within vendor limits |
| Compliance certifications | None (see limitations) | SOC 2, GDPR programs, enterprise SLAs |
Honest constraints — read before adopting. We'd rather you know these up front than discover them later. Most are the deliberate tradeoff of a free tool that runs entirely inside your own Google account.
Platform limits (Google Apps Script)
- Each action runs in Google's serverless sandbox with a 6-minute execution cap (30 minutes on Workspace) — very large bulk operations (mass scoring, big backups) can hit it.
- Daily quotas apply: outbound API calls, email sends (~100/day on consumer Gmail, ~1,500 on Workspace), and total trigger runtime. Heavy AI or email volume can bump these.
- Apps Script serializes execution, so many people acting at the same instant may queue briefly (guarded with locking — reliable, but not a high-concurrency backend).
- The public web app shows Google's small "created by a Google Apps Script user" banner (only removable on a paid Workspace domain).
Scale (Google Sheets as the database)
- Best for small-to-mid pipelines — hundreds to low-thousands of active candidates. Sheets slow down and hit cell limits well before enterprise volumes (the built-in archiver helps).
- It's a spreadsheet, not a real database — no true transactions; concurrency is guarded with locks, not an RDBMS.
AI (bring your own Gemini key)
- Not truly zero-cost: the software is free, but you use your own Gemini API key and pay pay-as-you-go (cents for small teams, but not $0).
- AI actions take a few seconds (parse a CV, score, draft an email) — assistive, not instant.
- AI is probabilistic — it can misjudge; it assists, it doesn't replace human review.
- Use a paid AI tier for real candidate data — free tiers may train on your inputs (a privacy call-out, not just cost).
Security & compliance
- Not SOC 2 / ISO 27001 certified, not third-party pen-tested — not enterprise-compliance-ready out of the box (regulated buyers like BFSI will need more).
- Your data lives in your own Google account — security depends on your account hygiene (2FA, careful sharing). Access control and audit are app-level: solid, but not infrastructure-grade.
- Data residency = Google's (region tied to your account); GDPR / India DPDP obligations for candidate PII are the operator's responsibility.
Product scope
- No built-in job-board integrations (LinkedIn/Naukri/Indeed) — intake is the public careers page, CV-by-email, or manual.
- No SSO/SCIM, no native mobile app (responsive web only); scheduling is basic (work hours + slots, not full calendar round-robin).
- Single company per install (single-tenant) — not a multi-tenant SaaS.
Operations
- Community open-source = no SLA, no guaranteed support or uptime. You self-host and self-support.
- Upgrades are manual (re-paste updated files); backups / disaster recovery are your responsibility (a built-in Sheet backup helps).
In one line: the most capable ATS you can run for free inside your own Google account, with real AI — built for startups and lean TA teams, not regulated enterprises. You bring a Gemini key and accept a few seconds of AI latency and Google's platform limits in exchange for owning your data and paying nothing for software.
If you're a 2,000-person company with a compliance team, buy an enterprise ATS. If you're a founder, an agency, or a TA team of one to fifteen who wants full control and a $0 software bill — this was built for you. (Scale, latency, compliance, and integrations are exactly what a database-backed successor addresses — see Optional modules.)
AgentATS is fully self-provisioning — you never touch a Sheet ID or folder ID:
- Go to script.google.com → New project.
- Paste in the six files from this repo (
Code.gs,TalentRubric.gs,Index.html,Apply.html,Source.html, and optionallyCvForwarder.gsfor the careers mailbox). - In Project Settings → Script Properties, add one property:
GEMINI_KEY= your Google AI Studio API key (free to create at aistudio.google.com). - In the editor, select
firstRunand click Run (authorize when asked). It creates your tracker spreadsheet and CV folder, builds every tab, makes you the Admin, and logs your new Sheet's URL. - Deploy → New deployment → Web app — Execute as: Me, Who has access: Anyone. Copy the URL.
That's it. Open the URL: that's your ATS. Add ?page=apply for your public careers page. Full click-by-click instructions (including a zero-code path for non-technical folks) are in SETUP.md.
- All data stays in the installer's own Google account — the spreadsheet, the CV files, the logs. Nothing is sent anywhere except your own Gemini API calls to Google.
- No secrets in code. API keys and webhook secrets live only in Script Properties. This repository contains property names and instructions — never values.
- Server-side, role-based access control with per-teammate access tokens; visitor identity is never trusted from the browser.
- Hardened intake: webhook authentication, daily submission caps, file-size limits, and formula-injection sanitization on everything written to the sheet.
- Audit trail of who changed what, and automated backups.
CvForwarder.gs— runs in yourcareers@mailbox and forwards every emailed CV into the ATS. Two clearly marked placeholders to fill in; see SETUP.md.- External database sync — a Supabase sync/backup module exists for teams that outgrow Sheets, but it is intentionally not part of the open-source core. The OSS release is Google-Sheets-only: one account, one mental model, nothing else to secure.
This project is maintained for the HR/TA community, by the HR/TA community. Bug reports, feature requests, translations, rubric improvements, and documentation fixes are all welcome — see CONTRIBUTING.md. If AgentATS saved your team money, the best thank-you is a contribution (or telling another recruiter it exists).
MIT — free for commercial and personal use. See LICENSE.