Skip to main content

Free Security Questionnaire Assistant for Matching Questions to Your Evidence

Keep an evidence library, paste a customer's questions and record your own answers with an owner and a status. It matches by keyword in your browser and never writes an answer.

Last updated 15 November 2026

Loading tool…

This security questionnaire assistant is a workspace for a business that has been sent a customer's security questionnaire and has to answer it from what it can show. You list the evidence you hold by typing a title, the statement each item supports in your own words, its date, its owner and where it is kept. You paste the customer's questions, and the page sorts each one into topics with a keyword table you can open and read. Against each question you pick an answer type, write the answer yourself, name an owner, set a status and link the evidence behind it. The page flags a Yes or Partly answer with no evidence linked, Yes and No answers on one topic and linked evidence older than an age you set, then exports a table for the person who reviews the answers. It runs in your browser, calls no AI service and never writes an answer for you. Nothing you type or paste is sent to a server. When you add questions, the page sends one analytics event that names the tool and the page.

How to use this tool

  1. 1

    Build your evidence library

    Add up to 200 items. For each one, type a title, the statement it supports in your own words, its kind, its date, its owner and where it is kept, tick the topics it supports and add up to 8 keywords of your own. Nothing is uploaded and the tool never reads a document.

  2. 2

    Paste the customer's questions

    Paste the questions, choose one question per line or by leading numbers, and press Add questions. One paste can be up to 60,000 characters and the list holds up to 300 questions. Load sample questions puts ten invented questions in the box to try.

  3. 3

    Record your answers

    Each question shows the topics its keywords matched and the library items that share a topic or a keyword. Pick an answer type, write the answer yourself, name an owner, set a status and link the evidence. Insert my statement into the answer adds your own evidence statement word for word.

  4. 4

    Read the flags

    The page lists its own checks: a Yes or Partly with no evidence linked, a Yes on a certification or testing question with no certificate or report linked, Yes and No answers on one topic, linked evidence older than the age you set, an evidence date after today and a question that matched no topic.

  5. 5

    Export for review

    Download the table as a CSV file or copy it as Markdown, and give it to the person who reviews the answers. Nothing is stored unless you press Save on this device, and Clear saved workspace removes what was stored.

How the keyword matching works, and where it stops

The page sorts questions with a keyword table of eight topics. Seven are the question groups of our vendor security questionnaire template: governance, application testing, data hosting, sub-processors, access control, incidents and resilience. The eighth, certifications and audit reports, is that template's row G3 on its own, because one of the checks treats it differently. The table is this page's own list, not a standard, and you can open it under the question box. A keyword is one or more words. The question and the keyword are both reduced to lower-case words, with every character that is not a letter or a digit read as a space, and the keyword matches when its words appear in the question in a row as whole words. A keyword that ends in an asterisk also matches longer words that start with its last word, so encrypt* matches encrypted and encryption, while sso matches only that word. A question can match several topics or none. An evidence item is suggested for a question when one of its topics is a topic the question matched, or when one of your own keywords for the item appears in the question. Items matched by one of your keywords are listed first, a question lists up to eight suggestions, and a picker under them reaches every item in the library. That is all the matching does. It does not understand a question. "Can a stolen credential alone open an account?" is about sign-in, but it holds no keyword from the table, so it matches no topic. "Will you notify us of price changes?" matches the incidents topic because of the word notify. Read each question yourself, link the evidence that fits whatever the suggestions say, and add keywords of your own to an item when your customers use a word the table lacks.

Why the tool never writes an answer

Every answer in a security questionnaire is a statement your business may later be asked to support. A sentence saying you hold a certification, have been tested or comply with a standard is only worth writing when it is true and you can show it, and a tool that cannot see your systems or read your documents has no way to know. So this one holds no stock answers and no wording of that kind. A new question starts as Not answered with an empty answer box, and nothing fills it in for you. The one piece of help is Insert my statement into the answer, shown beside each linked evidence item that has a statement. It adds the statement you wrote for that item, word for word, on a new line after anything already in the box, followed by a marker in square brackets that names the item, such as [Evidence: Access review record]. The marker lets a reviewer trace the sentence to its source and delete the marker before the answer is sent. If the answer would pass 2,000 characters, nothing is inserted and the page says so. For each question you also choose an answer type (Yes, No, Partly, Not applicable or Not answered), an owner of up to 80 characters and a status: Not started, Draft, Needs evidence, Ready for review or Reviewed. Write each statement in your library the way you would be happy to see it quoted back: what is true, for which system and as of which date. Where the honest answer is Partly, say where it stops. Our guide to answering security questionnaires covers how to word an answer that is not yet a yes.

What each flag means, and what none of them can tell you

The flags are this page's own checks on what you entered. None comes from a standard and none is a pass. Five are marked Review and one is marked Note. A Yes or Partly with no evidence linked is flagged, because the answer claims something and nothing in the workspace supports it. A Yes on a question matched to the certifications or application testing topics is flagged when no linked item has the kind Certificate or Report. This check does not look at Partly, and it reads only the kind you chose for the item. Yes and No answers on one topic are flagged once for that topic, with up to ten question numbers on each side. Only Yes and No count as opposite, and both can be right, so the flag asks for a read and not a change. Linked evidence dated more than a set number of months before today is flagged. You set the number, from 1 to 120. It starts at 12, which is this page's own starting value and not a rule from any standard, and today is the date on your device. An item with no date is not checked for age, and neither is an item that is not linked to an answer. An evidence date after today is flagged for any item in the library, linked or not. A question that matched no topic gets the Note. Each question lists the flags that name it, and the full list shows the first 100. An empty list means only that these six checks found nothing. The page cannot tell whether an answer is true, whether a document says what its title claims, or whether an answer will satisfy the customer who sent the questionnaire. A person reviews every answer before it is sent.

What the export holds, and how cells are escaped

The export is one table, as a CSV file or as Markdown to copy, with a row for each question and seven columns: the question, the answer type, your answer, the owner, the status, the titles of the linked evidence and where each item is kept. The answers in it are the text in your answer boxes. The tool adds the column headings, the labels of the choices you picked and the words no location recorded where an item has none. Questions are pasted from someone else's document, so a cell may begin with a character a spreadsheet treats as the start of a formula. OWASP's CSV Injection page says that when a spreadsheet program such as Microsoft Excel or LibreOffice Calc is used to open a CSV, any cells starting with = will be interpreted by the software as a formula. To remediate it, the page says to ensure that no cells begin with an equals sign, a plus sign, a minus sign, an at sign, a tab, a carriage return or a line feed, or with full-width variants of formula-initiating characters. It adds that this is not sufficient on its own: the field separator and quotes need taking care of too. In the CSV, a cell that begins with one of those seven characters, or with the full-width equals, plus, minus or at sign, gets a single quote in front, and every cell is wrapped in double quotes with a double quote inside it written twice. The Markdown table has no quoting. There, a tab or a line break inside a cell becomes a space, spaces at each end are dropped, the same single quote rule then applies, and a pipe or a backslash gets a backslash in front. The same OWASP page says wrapping each field in double quotes, prepending a single quote and doubling each double quote are not reliable in Microsoft Excel after saving and re-opening the CSV file, and that there is no universal CSV sanitisation strategy that is safe for all spreadsheet applications and all downstream consumers.

Saving on this device, and what leaves the page

Nothing you type is stored unless you press Save on this device. Each press writes one record to this browser's local storage under a key that names this tool. The record holds the workspace as it was at that moment, the day it was saved and a version number, and it can be up to 2,000,000 characters. A larger workspace is not saved, and the page says so. Changes made after a save are not stored until you press the button again, and the status line tells you when there are any. When the page is next opened in the same browser, it loads the saved record. A record that cannot be read whole, for example one written by a different version of the tool, is not loaded and stays where it is until you clear it or save over it. Clear saved workspace removes the record and leaves what is on screen. If the browser refuses to store the record, the page keeps the copy in memory for that visit and tells you it will be gone when the page is closed or reloaded. A shared computer keeps what you save. MDN says localStorage data has no expiration time and is saved across browser sessions. This page loads the saved record whenever it is opened in that browser, so the next person to open it there would see your questions, answers and evidence list. Clear the saved workspace before you leave, or do not save there. MDN adds that localStorage data for a document loaded in a private browsing or incognito session is cleared when the last private tab is closed. Nothing you type or paste is sent anywhere. Questions, answers and evidence are not uploaded, and the tool takes no file. When you add questions, the page sends one analytics event that names the tool and the page, and nothing you typed or pasted.

Sources

Worked example

An invented example, not taken from any business: an operations lead has one evidence item in the library, an access review record with the kind Configuration record and the topic Access control, and answers three questions.

Question 1, answered Yes, record linked
Is multi-factor authentication enforced for all staff?
Question 2, answered No
Are passwords shared between staff for any system?
Question 3, answered Yes, nothing linked
When was the application last penetration tested, and by whom?
Flags listed
3, all marked Review

The page flags Yes and No answers on the topic Access control for questions 1 and 2, and flags question 3 twice: Yes with no evidence linked, and Yes with no certificate or report linked.

The first flag needs only a read. Enforcing multi-factor sign-in and not sharing passwords agree with each other, and the page cannot tell. The other two stay until the test report is added to the library with the kind Report and linked to question 3, or the answer is changed to what the business can show.

Who uses this tool

Suppliers answering a customer questionnaire

Sort a long list of questions into topics, see which ones your evidence already covers and which need someone to go and find something.

Operations leads coordinating answers

Give each question an owner and a status, then export the table so a second person can review every answer before it goes back.

Founders preparing before one arrives

Build the evidence library first, with dates and where each item is kept, so the gaps show before a customer asks.

Teams reusing last year's answers

Paste the new questionnaire, link the same evidence and let the age check point out items nobody has refreshed.

Frequently asked questions

Does this security questionnaire assistant use AI to write answers?

No. It calls no AI service and does not understand a question. Nothing you type or paste is sent to a server. When you add questions, the page sends one analytics event that names the tool and the page. It compares the words in each question with a keyword table you can open on the page. You write every answer. Its one piece of help is adding your own evidence statement to an answer, word for word, with a marker naming the evidence item.

Where do my questions and evidence go?

They stay in the page. Nothing you type or paste is uploaded, and no document is read: you describe each evidence item in text. When you add questions, the page sends one analytics event that names the tool and the page, and nothing you typed or pasted. A copy is kept in this browser only if you press Save on this device.

What does Save on this device store, and how do I remove it?

One record in this browser's local storage: the workspace as it was when you pressed the button, the day it was saved and a version number. The page loads it the next time it is opened in the same browser, so on a shared computer the next person would see it. Clear saved workspace removes the record. If the browser refuses to store it, the copy is held in memory until the page closes, and the page tells you.

Why did a question match the wrong topic, or no topic at all?

Matching is by keyword only. A question worded without any keyword from the table matches no topic, and a question that shares a word with another subject can pick up a topic it is not about. Each question lists the topics it matched and the keywords that matched them, and the keyword table on the page shows the full list. You can link any item in your library to any question whatever the match says, and add keywords of your own to an evidence item.

Does a workspace with no flags mean the answers are ready to send?

No. The flags are this page's own checks on what you entered, such as a Yes with no evidence linked. None can tell whether an answer is true, whether a document says what its title claims, or whether the answer will satisfy the customer who sent the questionnaire. A person reviews every answer before it is sent.

What does the export do about spreadsheet formulas?

A cell that begins with a character such as =, +, - or @ gets a single quote in front, and every CSV cell is wrapped in double quotes, with quotes inside doubled. OWASP's CSV Injection page says to ensure that no cells begin with those characters and to take care of the field separator and quotes. It adds that wrapping fields in double quotes, prepending a single quote and doubling double quotes are not reliable in Microsoft Excel after saving and re-opening the CSV file.

Related tools and services