RFxCheck

Open source semantic requirement checking for procurement. Highlights RFx requirements, deadlines and conditions as you read.

RFxCheck helps you spot important procurement requirements while you read an RFx webpage. Instead of generating a summary, RFxCheck applies seven repeatable semantic checks to the page and highlights the original source clauses that match. The clause stays the evidence: RFxCheck adds an outline and a category badge, and never rewrites, replaces or reorders page content. WHAT RFXCHECK CHECKS FOR • Mandatory requirements — obligations a bidder, supplier, product or submission is expected to satisfy • Deadlines — closing dates, submission deadlines and other time-sensitive requirements • Disqualification conditions — conditions that may make a response non-compliant, ineligible or rejected • Insurance and bonding — insurance, bonding, surety and related financial protection requirements • Certifications and licences — licences, certifications, accreditations, permits and formal qualifications • Mandatory events — required site visits, bidder meetings, demonstrations, interviews or inspections • Submission instructions — how, where, when or in what format a response must be submitted A clause often matches more than one check, and RFxCheck shows a badge for each. HOW IT WORKS 1. Open an RFx webpage. 2. Open RFxCheck from the toolbar. 3. Select Scan This Page. 4. RFxCheck reads the meaningful, visible text of that page and evaluates each section against the seven checks. 5. Matching clauses are highlighted in place, with a badge for every check that matched. Select Clear Highlights to remove everything RFxCheck added. BRING YOUR OWN KEY RFxCheck is powered by TypeSafe Jev and uses a bring-your-own-key model. You supply your own TypeSafe Jev API key, and it is stored locally in your own Chrome profile. When you start a scan, relevant readable text from that page is sent directly from your browser to the TypeSafe Jev API for semantic classification. RFxCheck does not send raw HTML, the page URL, cookies, form contents or images. Text sent to TypeSafe is handled under TypeSafe's own terms and privacy policies. Please read them, and your own organisation's policies, before scanning anything confidential. PRIVACY • No RFxCheck backend. There is no RFxCheck account, no RFxCheck server, and no developer-owned API key anywhere in the extension. • No analytics, no telemetry, no tracking. • Your API key is stored locally, is used only to authenticate directly to TypeSafe, and is never sent to the developer of RFxCheck. • A page is read only when you invoke RFxCheck on that tab. RFxCheck declares no content scripts and requests no broad host access. Full policy: https://github.com/NeerajMohanty/RFxCheck/blob/main/PRIVACY.md OPEN SOURCE RFxCheck is open source under the Apache License 2.0. The seven checks, the 0.80 decision threshold and the entire request the extension sends are readable in the repository. https://github.com/NeerajMohanty/RFxCheck LIMITATIONS, STATED PLAINLY • Normal webpages only. There is no PDF parsing, no OCR and no file upload. • At most 75 meaningful text blocks per scan, so a very long solicitation may be partially covered. The popup tells you when that limit was reached. • Sections are evaluated independently, without surrounding context, so a table cell may be judged without the date in the next cell. • Results close to the threshold may vary between runs, because Jev returns a calibrated probability rather than a fixed label. • Semantic classification can be wrong in both directions. Findings are semantic classifications intended for human review. RFxCheck does not determine bid compliance and is not legal advice. RFxCheck is an independent open source project. It is not an official TypeSafe product and is not affiliated with, endorsed by or sponsored by TypeSafe.