Free Pentest Scope Estimator for Web Applications
List the systems you own or hold written permission to test and get a target register, an exclusions list and an effort range in person-days. Your answers stay in your browser.
Last updated 15 November 2026
This pentest scope estimator is a planning form for one web application. You describe the environments on offer, the roles that sign in, the sign-in features, the number of screens or routes and API endpoints, and features such as file uploads, payments, an admin area, more than one tenant, webhooks and source code access for the tester. Then you list each target and tick it only if you own it or hold written permission to test it. Press Build the scope and the tool arranges your answers into three outputs: a target register that keeps authorised targets apart from those that still need permission, an exclusions list that starts from the general exclusions in the scope template in this library, and an effort range in person-days with every line of its arithmetic on screen. It checks, looks up and requests nothing, and it shows no price. The guide below explains how to read each output and where the tool stops.
How to use this tool
- 1
Describe the application
Add an optional label, tick whether a staging copy is offered and whether it has seeded test data and test accounts, say what it shares with production, tick whether a web application firewall or rate limit sits in front, choose what testing production allows, and note any testing window. List the roles one per line, up to 12, tick the sign-in features and other features, and enter the two counts, each from 0 to 2,000.
- 2
List the targets
Press Add a target for each system, up to 40. Type the target, choose its environment and who runs it, name its owner, and tick "We own this or hold written permission to test it" only where that is true. Add up to 10 exclusions of your own, each with a reason code.
- 3
Build the scope
Press Build the scope. If an answer is missing, repeated or over a limit, or two answers would list one target as both authorised and excluded, the tool lists the problems, up to 12 at a time, and builds nothing until they are fixed. Otherwise it shows the target register, the exclusions list and the effort range. Changing an answer afterwards removes the outputs until you build again.
- 4
Adjust the weights and take the outputs away
Change any weight in the effort table, or press Reset the weights. Copy each output as Markdown, download the target register as CSV, and press Save on this device if you want the answers and the weights kept in this browser.
What a penetration testing scope has to settle before anyone quotes
A penetration testing scope is the written answer to three questions: what may be tested, what must be left alone, and how much there is to test. Until those are settled, a quote is a guess and a test has no edge. This form asks for the answers in the order they are easiest to give. Environments come first, because where testing runs changes what a tester may do: a staging copy with seeded test data and test accounts, and nothing shared with production, can be written to freely, while production may allow nothing, only checks that need no account and change nothing, or signed-in testing with test accounts only. Roles come next, because access rules are tested between people, and each role you list adds accounts to prepare and comparisons to make. Sign-in features, file uploads, payments, an admin area, more than one tenant and webhooks each bring checks of their own, so each is a separate answer. The two counts, screens or routes and API endpoints, describe size. Then come the targets themselves, one row for each system the test would send requests to. Before you start, gather what the form will ask for, as a short pentest scope checklist: the hostnames and API addresses with the environment each belongs to, the roles that can sign in, what a staging copy shares with production, a count of screens or routes and of API endpoints taken from the code or the API schema, the outside services the application relies on and who runs each, and any limit on when testing may happen. The scope is one of three documents. The scope says what will be tested. An authorisation letter, signed by a person with authority over each target, gives permission to test it. Rules of engagement govern conduct inside the testing window. This tool drafts only the first, and its outputs are meant to be carried into the other two.
How the target register keeps authorised targets apart from the rest
The target register is an editable list of up to 40 rows. Each row holds the target as text, its environment (production, staging or other), who runs it, its owner and one tick box: "We own this or hold written permission to test it". The tick is the only thing that puts a target in the list headed Authorised targets. A row without it goes one of two ways. If your business runs the target, it is listed under Needs permission before any testing, where it stays until you can truthfully tick it. If another party runs it, such as a payment provider, a hosted sign-in service or a host's shared platform, it is listed under Excluded: run by another party and gets a line in the exclusions list with reason code O, owned by someone else. The tool states nothing about what any named provider allows. It says only that the provider's own policy decides, because a business can authorise testing only of what it controls. Above the lists, the register summarises your other answers, including the number of test accounts to prepare: two for each role, in each of two organisations when you ticked more than one tenant, which is what the scope template in this library asks for. Notes can appear under the lists. One gives the scope template's warning that a staging copy sharing production's database or file storage carries production's risks, when yours shares something or you have not said. One reminds you to attach a party's written permission, or its published testing terms with the date you read them, when a ticked target is run by another party. One tells you that an authorised target is in staging while no staging copy is ticked. A ticked target in production while testing allowed in production is None stops the build instead. The tool cannot check any of this. It does not look up a hostname, test whether it answers or confirm who owns it, so the register is exactly as accurate as what you typed and ticked.
The exclusions list: four lines from the scope template, then your own
The list this tool builds starts with four general lines taken, word for word, from the completed example in the scope template in this library: the vendor and host platforms in Part 6, load and denial-of-service testing, staff laptops, the office network and phishing, and features released after the schedule is signed. Part 6 is that schedule's list of third-party services and hosting. Every line carries one of the template's four reason codes: O, owned by someone else; L, adds effort, to be scoped later; S, needs a different kind of specialist; R, too risky in the environment offered. Your answers then add lines. If testing allowed in production is None, the list adds all testing in production with code R, and if it is limited to checks that need no account and change nothing, it adds signed-in testing in production. If you ticked payments, it adds the payment provider's side with code O, in the template's words: the hosted checkout page, card handling and the provider's dashboard. Its note lists what the template says is yours, tested in the provider's test mode, and says the line covers hosted card fields you ticked. If you ticked single sign-on, it adds the single sign-on provider's hosted login pages and platform, where another party runs them. Each unticked target that another party runs gets a line of its own, naming the target, the kind of party and the owner you typed. The form will not build while a ticked target is in production and testing allowed in production is None, or while a line you added is a ticked target. A line about a ticked target's party, and a target or added line on its host or with its text in another letter case, names that target as not covered. A line you word yourself is not read for meaning, so check it against the register. Last come up to 10 lines of your own. The four template lines cannot be removed in the tool, so edit the copied Markdown if your scope differs. Its Decided by column is left blank.
How the effort range is worked out, and what it is not
The effort range is this page's own arithmetic, and all of it is on screen. There are sixteen lines. One is counted on every scope, for planning, set-up and reporting. Four are counts: screens or routes, API endpoints, roles and the number of sign-in features ticked. Ten are yes or no answers that count as 1 or 0: file uploads, payments, an admin area, more than one tenant, webhooks, source code access for the tester, some testing allowed in production, no staging copy that is seeded and shares nothing with production, a web application firewall or rate limit in front, and a stated testing window. The last line counts authorised targets after the first, so a target without the tick adds nothing. Each line has a low weight and a high weight, in person-days for one unit of that line. The tool multiplies the count by each weight to get two subtotals, adds the low subtotals and the high subtotals, then rounds the low sum down and the high sum up to a half day. The result is always written as a range, low to high. No standard and no provider states these weights. They are starting values chosen for this page, so every weight is a field you can change, from 0 to 20 with at most two decimal places, and Reset the weights puts the defaults back. A weight the tool cannot read, or a low weight above its high weight, removes the range until it is fixed. Treat the result as a planning figure for your own conversations with providers. It is not a quote or a fee, it contains no price, and it is not a statement of how long any provider, Clever Ops included, will take. A tester who reads your route list and works through your application will learn things this form cannot ask about, and the estimate is only as good as the counts you typed.
What stays in your browser, and what Save on this device stores
Everything you type, tick and select stays in this page. The tool makes no request with any of it, and no target is looked up or contacted. Each time a scope is built, the page sends one analytics event that names the tool and the page, and nothing you entered. The tool stores none of your answers unless you press Save on this device. That writes one versioned record, holding your answers and the weights, to this browser's local storage under the key cleverops.pentest-scope-estimator.saved-scope. A record that would be longer than 40,000 characters is not saved, and the tool says so. MDN says data in local storage is saved across browser sessions and has no expiration time, and that local storage data for a document loaded in a private browsing or incognito session is cleared when the last private tab is closed. Clear saved scope removes the record, and Load saved scope puts the saved copy back into the form, replacing what is there. When the tool opens and finds a saved record, it loads it into the form and tells you. A record it cannot read, because it is damaged, too large or from a different version of the tool, is not loaded, and the tool says so and leaves it in place until you clear it. Storage can also refuse. MDN lists a SecurityError that reading the localStorage property can throw, and one of its cases is a request that violates a policy decision, for example when the user has configured the browser to prevent the page from persisting data. The tool catches storage errors, keeps the copy in this page's memory instead, and tells you that copy will be gone when you close or reload the page. Keep in mind who else uses the browser profile you save in: the record is your list of systems, their owners and what is excluded, and this tool does not encrypt it.
Copying the outputs, and what the CSV file does to each cell
Each of the three outputs has a copy block that holds it as Markdown. In the Markdown, the tool puts a backslash in front of each backslash, backtick, asterisk, underscore, square bracket, angle bracket and pipe in anything you typed, so a pipe in a target does not end a table cell. A table cell that begins with =, +, -, @ or a full-width form of one of them also gets a single quote in front, as in the CSV. The target register also downloads as a CSV file, pentest-target-register.csv, built in the page with nothing uploaded, holding every target and the list it is in. OWASP's CSV Injection page says that when a spreadsheet program opens a CSV, any cells starting with = will be interpreted by the software as a formula. To remediate this, it says to ensure that no cells begin with =, +, -, @, a tab, a carriage return, a line feed or a full-width variant of a formula-initiating character. It adds that this is not sufficient: the field separator, such as a comma or semicolon, and quotes need care too, because an attacker could use them to start a new cell with one of those characters at its beginning. So the tool turns tabs and line breaks into spaces and trims each entry, puts a single quote in front of a cell that then begins with =, +, -, @ or the full-width form of one of those four, wraps every cell in double quotes and doubles each double quote inside it. The same page says double quotes with a single quote in front are not reliable in Microsoft Excel after saving and re-opening the CSV file: Excel may remove quotes or escape characters, and previously escaped formulas may become active again. It says there is no universal CSV sanitisation strategy that is safe for all spreadsheet applications and all downstream consumers. It also describes a tab prefix as an Excel-resistant mitigation, which this tool does not add. Read the file as a list of text you typed, not as checked data.
Sources
- OWASP Foundation: CSV Injection (OWASP community attacks page)
- MDN Web Docs: Window: localStorage property
- Internet Engineering Task Force (IETF), RFC Editor: RFC 3986: Uniform Resource Identifier (URI): Generic Syntax
Worked example
An invented example, not a client: a trades business scopes its customer portal before asking for quotes. The hostnames it types end in .lab.test and are invented for this example.
- Staging copy
- Offered, with seeded test data and test accounts, sharing nothing with production
- Testing allowed in production
- None
- Roles
- Customer, Field staff, Office administrator
- Sign-in features
- Password sign-in, Password reset
- Size
- 40 screens or routes, 0 API endpoints
- Features
- File uploads, payments with hosted card fields, webhooks, source code access
- Targets
- Staging portal, ticked. Production portal, not ticked. Payment provider's checkout page, not ticked
One authorised target, one under Needs permission before any testing, one excluded, seven exclusion lines, and an effort range of 5 to 11.5 person-days with the default weights.
The sums behind the range are 5.05 and 11.25 person-days, rounded outwards to half days. The register lists 6 test accounts to prepare, two for each role. The production portal stays out of the authorised list until someone can tick it, and the provider's page became an exclusion with reason code O. The range moves with any weight the business edits, and it says nothing about what a provider will quote.
Who uses this tool
Business owners about to ask for quotes
Write down what you run, who runs each part and what must be left alone, then send the same register and exclusions list to every provider you approach.
Developers and agencies briefing a tester
Turn the route count, the roles and the environments into a register a tester can read in a minute, with the systems that still need a signature listed apart.
Teams answering a customer who asked for a test
Draft the scope before the scoping call, so the call starts from a list of targets and exclusions and not from a blank page.
Buyers comparing quotes
Hold each quote against your register: the same targets, the same roles and the same exclusions, with the effort each provider states beside your own planning range.
Frequently asked questions
Does the pentest scope estimator send my answers anywhere?
No. Everything you type, tick and select stays in the page, and no target is looked up, requested or contacted. Each time a scope is built, the page sends one analytics event that names the tool and the page, and nothing you entered. The tool stores none of your answers unless you press Save on this device, and Clear saved scope removes that record.
Is the effort range a quote?
No. It is arithmetic on the counts you typed and the weights shown, written as a low to high range in person-days. The weights are this page's own starting values, not figures from a standard or a provider, and you can change every one. The range is not a fee, and it is not a statement of how long any provider, Clever Ops included, will take. Clever Ops quotes each test after a scoping call instead of publishing prices.
What does ticking "We own this or hold written permission to test it" do?
It is the only thing that lists a target as authorised. Without the tick, a target your business runs is listed under Needs permission before any testing, and a target another party runs goes to the exclusions list. The tick records your own statement. The tool cannot confirm that you own a system or hold permission to test it, so tick a row only when it is true.
Why is my payment provider or hosted sign-in service in the exclusions list?
Because another party runs it. A business can authorise testing only of what it controls, so the tool excludes another party's system by default with reason code O, owned by someone else. It makes no statement about what any named provider allows: the provider's own policy decides. If you hold that provider's written permission, tick its row in the target list and attach the permission to your scope. Each line about that provider stays in the exclusions list and says it does not cover that target.
Can I use this as a penetration testing scoping questionnaire?
Yes, as one you fill in for yourself before you ask for quotes. It asks for environments, roles, sign-in features, sizes, features and targets, and it returns them as a register, an exclusions list and an effort range you can send with your request. A provider may still ask its own questions, and the tool does not write the authorisation letter or the rules of engagement.
How long does a saved scope stay on my device?
Pressing Clear saved scope removes it. Until then, MDN says data in local storage is saved across browser sessions and has no expiration time, and that local storage data for a document loaded in a private browsing or incognito session is cleared when the last private tab is closed. The record holds your answers and the weights, under one key that names this tool.