Vulnerability disclosure policy template for Australian businesses
A policy you can copy and adapt that tells security researchers which systems they may look at, how to report what they find and what your business will do next, with a security.txt checklist and one report followed from first email to closure.
It is general information, not legal advice: your organisation approves its own scope, contacts and terms, and its solicitor should review the safe-harbour and privacy wording before it goes live.
On this page
Copy-ready disclosure policy, security.txt checklist and worked report
A vulnerability disclosure policy to copy and adapt, with a prompt in round brackets for every decision your organisation makes, a security.txt publishing checklist keyed to RFC 9116 sections, and one invented researcher report followed from first email to closure. It is general information, not legal advice: your organisation approves its own scope, contacts and terms, and its solicitor reviews the safe-harbour and privacy sections before publication.
vulnerability-disclosure-policy-template.txt
GENERAL INFORMATION, NOT LEGAL ADVICE. This is a starting point, not a finished policy. (Organisation name) decides and approves its own scope, contacts, response targets and disclosure terms, and its solicitor should review sections 4 and 8 before anything is published. Text in round brackets is a prompt: replace it with your own words or delete it. Lines that start with INTERNAL NOTE are for you and come out before publication. Three parts follow: Part A. The policy, to adapt and publish as a page on your website. Part B. A security.txt publishing checklist based on RFC 9116. Part C. A worked example: one invented report followed from first email to closure. # PART A. THE POLICY # (Organisation name) vulnerability disclosure policy Version (number). Published (date). Next review due (date). ## 1. Why this policy exists We want to hear from anyone who finds a security weakness in our systems. This policy explains which systems it covers, what we ask of you while you look, how to report what you find, and what we will do with your report. If you think you have found something, please tell us even if you are unsure it matters. A short report is better than none. ## 2. Systems this policy covers In scope: - (Main website address, for example www.yourbusiness.com.au) - (Customer portal or web application address) - (API address, and the versions in use) - (Any other system you are willing to receive research on, or delete this line) Out of scope. Please do not test: - Any system not listed above, including other subdomains of our domains. - Services other businesses run for us, including (hosted payment pages, booking platform, email service, accounting software and any other supplier system). - Our premises, staff and customers. No physical access attempts, and no phishing, phone calls or other social engineering. INTERNAL NOTE: list a system as in scope only if you are permitted to allow research on it. Where a supplier runs the system, check your contract or the supplier's published policy first, and leave the system out if the supplier has not agreed. If you are not sure whether something is in scope, ask us at (security contact) before you test it. Weaknesses in products or services we use but do not run, such as a plugin, a code library or a hosted platform, belong with that supplier under its own disclosure policy. If you prefer, tell us and we will pass the report on. We will not share your name or contact details with the supplier without your permission. ## 3. What we ask of you While researching under this policy, please: - Test only the systems listed as in scope in section 2. - Use accounts you created yourself, or accounts whose holders have given you explicit permission. - Access only the minimum data needed to show the weakness exists. If a weakness exposes other people's data, stop and report it rather than exploring further. - Avoid harm: no denial-of-service or load testing, no changing or deleting data that is not yours, nothing that degrades the service for other users, and no using a weakness to reach other systems or keep ongoing access. - Keep any automated scanning at or below (the request rate your systems can tolerate), and stop if the service slows down. - Report promptly, and discuss the weakness with us only through the contact methods in section 5. - Keep the details confidential until we have agreed on disclosure under section 7. - Do not ask for payment as a condition of sharing details or of keeping them private. This policy does not offer payment for reports. If you come across personal information, follow section 8. ## 4. Our commitment if you follow this policy INTERNAL NOTE: this section states (Organisation name)'s own intent. Have your solicitor review it before publication. If you make a good-faith effort to follow this policy: - (Organisation name) will not take legal action against you over your research, and will not ask anyone else to. - If someone else takes legal action against you over research that followed this policy, we will make it known that your research followed this policy. - We will work with you to understand the weakness and fix it. What this commitment does not do: - It covers only decisions (Organisation name) controls. It does not bind police, regulators, our suppliers or anyone else. - It does not extend to systems outside the scope in section 2, or to anything section 3 asks you not to do. - It is not a statement that any activity is lawful, and it is not legal advice to you. INTERNAL NOTE: some published templates also state that the organisation considers good-faith research authorised. Whether to add a statement like that, and in what words, is for your solicitor to decide. If you are unsure whether something you plan to do fits this policy, ask us at (security contact) first. ## 5. How to report Send your report to (security email address, for example security@yourbusiness.com.au) or through (address of your reporting form). (If you accept encrypted email, give the address where your public key can be downloaded.) Please include: - The system, and the exact page, route or API endpoint involved. - What the weakness is, and what someone could do with it. - Steps we can follow to reproduce it, including the type of account you used. - Supporting evidence, such as requests and responses, screenshots or a short screen recording, with any personal information redacted. - When you tested, and the IP address you tested from, so we can tell your traffic apart from other activity in our logs. - Whether you have told anyone else, and any date on which you plan to publish or present your findings. - How you would like to be credited, if at all. You can report without giving your name. If you give contact details, we use them only to handle your report and will not publish them without your permission. Please write in (English, or the languages your team can handle). ## 6. What happens after you report These are targets (Organisation name) sets for itself, not promises: - Acknowledge your report: within (number) business days. - Tell you whether we could reproduce the weakness: within (number) business days of acknowledging it. - Update you while we work on it: at least every (number) days until the case is closed. - Aim to fix confirmed weaknesses according to our rating: Critical within (number) days, High within (number) days, Medium within (number) days, Low by (the next scheduled release, or a number of days). If we expect to miss a target, we will tell you why and give you a new date. Fixes that depend on a supplier can take longer. We may decide that a report does not describe a security weakness, or that we will accept a risk rather than change a system. If so, we will explain our reasoning to you. When a fix is in place we will tell you, and we may ask you to check it. You are under no obligation to. ## 7. Disclosure INTERNAL NOTE: choose Option 1 or Option 2 and delete the other. Option 1, an agreed date. Please keep the details private until we have agreed a disclosure date with you. We will raise it once we have confirmed the weakness, and we will aim for a date that gives our customers time to receive the fix. Option 2, a default window. Unless we agree a different date with you, please keep the details private for (number) days from the day we acknowledge your report, or until we tell you the fix is in place, whichever comes first. Under either option: - We may ask for more time if a fix is in progress, and we will explain why. - We may publish our own notice sooner if the weakness is being actively exploited, and we will tell you before we do. - When you publish, please leave out personal information, customer details and anything that would help someone attack a system that is not yet fixed. - (If you keep an acknowledgements page: with your permission, we will credit you by the name or handle you choose at (acknowledgements page address).) If the weakness affects other organisations that use the same product, we may report it to the supplier and, where the supplier cannot be reached, to (who you will tell, for example ASD's ACSC through the vulnerability option on its Submit a report page). We will not share your name without your permission. ## 8. Personal information If your research exposes personal information about anyone other than you: - Stop testing that weakness. - Do not download, copy, keep or share the information. - Report to us straight away. Describe what kind of information you saw and how, without sending us the information itself. - Delete anything you saved by accident, and tell us when you have. Information about you: we use your contact details and your report only to deal with your report, as set out in our privacy policy at (privacy policy address). INTERNAL NOTE, data breach assessment: treat a report showing that personal information was exposed as a possible data breach. Tell your privacy officer the same day and follow your data breach response plan. Under the Notifiable Data Breaches scheme, an organisation covered by the Privacy Act 1988 must notify affected individuals and the OAIC when a data breach is likely to result in serious harm to a person whose personal information is involved. Get advice on whether the scheme applies; do not leave that assessment to the person who reads the security inbox. ## 9. Questions and changes Send questions about this policy to (security contact). We review it at least every (period), and whenever our systems or contacts change. Change history: Version (number), (date): (what changed) # PART B. SECURITY.TXT PUBLISHING CHECKLIST Based on RFC 9116, "A File Format to Aid in Security Vulnerability Disclosure" (IETF, April 2022). Section numbers refer to the RFC. Before publishing [ ] Decide which hosts get a file. A security.txt file applies only to the domain or IP address it was retrieved from, not to its subdomains or parent domains (section 3.1). Hosts that need one: (list them). [ ] Choose the contact methods. At least one Contact field is required. Use mailto: for an email address, tel: for a phone number, or an https:// address for a web page or form, and list them in your order of preference (section 2.5.3). [ ] Publish the disclosure policy first, so the Policy field has a page to point to (section 2.5.7). [ ] Choose an Expires date and time less than a year away, as the RFC recommends (section 2.5.5), and diarise a reminder well before it. Writing the file [ ] Exactly one Expires line, written in the RFC 3339 date and time format, for example 2027-05-31T14:00:00Z (section 2.5.5). [ ] If you include Encryption, it points to where your key can be downloaded. The key itself does not go in the Encryption field (section 2.5.4). [ ] Add optional fields only as needed: Acknowledgments, Canonical, Preferred-Languages (once only) and Hiring (sections 2.5.1 to 2.5.8). [ ] If you sign the file with an OpenPGP cleartext signature, which the RFC recommends, also add a Canonical line listing the address the file is served from (sections 2.3 and 2.5.2). Serving the file [ ] Serve it at https://(your domain)/.well-known/security.txt (section 3). [ ] Serve it over HTTPS with a Content-Type of text/plain and the charset set to utf-8 (section 3). [ ] If you also serve /security.txt at the top level for older tools, redirect it to the /.well-known/ file rather than keeping a second copy. Where both exist, the /.well-known/ file is the one that must be used (section 3). [ ] On a shared or multi-tenant platform, reserve both security.txt names at the root so no one else can create them (section 5.6). After publishing [ ] Send a test report to every Contact address from outside the business, and confirm who receives it. [ ] Watch the file, and the pages it points to, for changes you did not make. An attacker who compromises the site can alter the file or redirect it (section 5.1). [ ] Before the Expires date, check every contact and link, then publish a new date. The RFC says having no file may be preferable to having stale information in one (section 5.3). [ ] Plan triage time for automated reports, which the RFC says become more likely once a file is published (section 5.8). [ ] Remember what the file does not do. Researchers should not read its presence as permission to test, or its absence as a refusal (section 5.5). Say what you permit in the policy, which is one of the places the RFC says permission may be indicated. Example file (replace each prompt in round brackets): # Security contact for (your domain) Contact: mailto:(security email address) Contact: https://(your domain)/(path to your reporting form) Expires: (a date and time less than a year away, for example 2027-05-31T14:00:00Z) Policy: https://(your domain)/(path to your disclosure policy) Preferred-Languages: en Canonical: https://(your domain)/.well-known/security.txt # PART C. WORKED EXAMPLE: ONE REPORT FROM INBOX TO CLOSE ILLUSTRATION ONLY. The business, researcher, systems, times and findings below were invented for this page. They are not a real case. The business: a building-supplies wholesaler with a trade portal at portal.wholesaler.example, where account customers view orders and download invoices. Its policy, adapted from Part A, sets these targets: acknowledge within 2 business days, confirm or close within 5 business days of acknowledging, update at least every 7 days, and fix High-rated weaknesses within 14 days. It chose Option 1 for disclosure. An outside agency maintains the portal. Monday 9:40 pm. The report arrives A researcher writing as "kestrel" emails security@wholesaler.example. While signed in to their own trade account, they changed the invoice number in an invoice download address, /invoices/(invoice number)/pdf, to the number before it, and received an invoice belonging to another trade customer. They stopped after that one request, opened no other invoice and kept no copy. The report includes the request and response with the other customer's details redacted, the time of the request, the IP address they tested from and a request to be credited as "kestrel". It names no publication date. Tuesday 8:30 am. Logged and acknowledged The office manager, who watches the security inbox, logs the report as VDP-004 and forwards it to the agency's developer and, because personal information may be involved, to the privacy officer. The reply to the researcher says the report has arrived, gives the reference and promises the next update by Thursday. Tuesday 2:00 pm. Reproduced on staging The developer creates two test trade accounts belonging to two test businesses on the staging copy. Signed in as the second, a request for the first business's invoice returns the PDF. The weakness is confirmed without opening another real customer's record. The cause: the download route checks that someone is signed in, but not that the invoice belongs to their business. Tuesday 4:30 pm. Rated and assigned Rating: High. Any signed-in trade customer could read any other customer's invoices, which carry business names, contact names, delivery addresses and order details. Owner: the agency's lead developer. Interim measure: invoice downloads are switched off in the portal, and invoices are emailed to customers on request. Wednesday. Logs and privacy The host keeps 30 days of access logs. The developer searches them for invoice requests from accounts that did not own the invoice. The only match is the researcher's single request, at the time and from the address in the report. The case record notes that activity older than 30 days cannot be checked. The privacy officer records the assessment, with the reasons and the lawyer's advice, in the breach register. Thursday. Fixed on staging, researcher updated The fix looks up each invoice by both its number and the signed-in customer's business, and returns the same not-found response for an invoice that does not exist and one that belongs to someone else. A regression test with two test businesses now runs on every deployment. The developer checks other routes built the same way and finds that statement downloads had the same flaw, which is fixed in the same release. The researcher gets the promised update: confirmed, rated High, fixed on staging, production release planned for Friday. Friday. Released and checked The fix goes to production and invoice downloads are switched back on. The developer repeats the test on production with the two test accounts: each business sees its own invoices and gets not-found for the other's. The researcher is told the fix is live and offered the chance to check it. They confirm that a changed invoice number now returns not-found. The following week. Disclosure and credit Under Option 1, the business and the researcher agree a date from which the researcher may publish a short write-up that names the type of flaw but not the portal's customers. With permission, "kestrel" is added to the acknowledgements page. Any notice to customers follows the privacy officer's assessment, not the disclosure agreement. Closing the case The record for VDP-004 holds: the report and evidence; the times of arrival, acknowledgement, reproduction, fix and release; the rating and owner; the interim measure; the log search and its 30-day limit; the privacy assessment reference; the second route found and fixed; the regression test; the agreed disclosure date; and the credit given. The agency adds an ownership check for every new download route to its review checklist. What the example shows - Acknowledging inside the published target kept the researcher dealing with the business, not counting down to a publication date. - Reproducing with test accounts confirmed the flaw without opening a second real customer's invoice. - Checking routes built the same way found a flaw the report did not mention. - Recording the 30-day log limit states what was checked and what could not be. - Keeping the privacy assessment separate from the disclosure agreement means each decision is made by the person responsible for it.
What a vulnerability disclosure policy is for
A vulnerability disclosure policy is a public page that tells people outside your business what to do when they find a security weakness in your systems: where to send it, what they may and may not do while looking, and what you will do once it arrives. Its companion is security.txt, a small text file that points anyone who goes looking to the right inbox.
Reports arrive whether or not a policy exists. A trade customer notices that changing a number in a web address shows someone else's invoice. A developer using your API gets back more data than they asked for. Someone runs a scan and emails the results to the address on your contact page.
Without a policy, those messages land with whoever reads that inbox, who has no way to tell a genuine report from a sales pitch or a threat and no instructions for what happens next.
OWASP's Vulnerability Disclosure Cheat Sheet makes the same point from the receiving side: the staff who monitor the main contact address, web chat and phone lines should know how to handle security reports and who to escalate them to.
The CERT Coordination Center (CERT/CC), which has coordinated the disclosure of software vulnerabilities since 1988, defines coordinated vulnerability disclosure as the process of gathering information from the people who find vulnerabilities, coordinating the sharing of that information between the relevant stakeholders, and disclosing the existence of the vulnerabilities and their mitigations to stakeholders, including the public.
FIRST's guidelines on multi-party coordination say every party, vendors in particular, should publish actionable disclosure policies and expectations, including timelines and thresholds for disclosure.
A coordinated vulnerability disclosure policy is how a business sets those expectations before its first report rather than in the middle of it.
This page and the template are general information, not legal advice. The organisation that publishes the policy decides and approves its own scope, contact details, response targets and disclosure terms, and its solicitor should review the safe-harbour and privacy sections before anything is published.
Policy, program or bug bounty
The three terms are often used as if they were one, but each commits a business to something different.
The policy is the published document. A vulnerability disclosure program is the work behind it: someone reading the inbox, a way to log and rate reports, a developer who can fix what is confirmed, and someone who keeps the researcher informed until the case is closed. A bug bounty adds rewards for eligible findings, which OWASP says are usually monetary and are often managed through a third-party platform.
OWASP is direct about the order. Its cheat sheet lists the challenges a bug bounty brings for smaller organisations, including the time and resources to respond to reports, staff skilled enough to triage them, large numbers of junk or false-positive submissions, and difficulty telling legitimate testing traffic from attacks.
It says bounties should only be used by organisations that already have a mature vulnerability disclosure process, supported by strong internal processes to resolve vulnerabilities. On that advice, a policy and a working program come before any bounty.
The template on this page sets up both, offers no payment for reports and says so plainly, because a policy that is silent on rewards invites the question at an awkward moment. If you want to thank researchers another way, CERT/CC's principles note that it is usually better to reward good behaviour than to try to punish bad behaviour, and OWASP lists non-cash options such as credit in a hall of fame or similar acknowledgement.
Decisions to make before you fill in the blanks
Every prompt in the template is a decision only your organisation can make.
CERT/CC's guide lists what a disclosure policy typically covers: scope, exceptions, safe harbour, report quality requirements, preferred languages, contact information and timing. Five decisions settle most of the wording.
List each domain, subdomain, application and API you are willing to receive research on, and name what is excluded. CISA's template, written for US federal agencies, adds a test worth applying anywhere: before adding a system, make sure you are permitted to authorise testing on it. Where a managed service provider or software-as-a-service vendor runs the system, confirm whether that vendor has explicitly authorised such testing, in your contract or its own published policy.
If it has not, the template says to work with the vendor to obtain authorisation, and to leave the system out if that is not possible.
For many businesses that rules out the hosted checkout page, the booking platform, the email service and the accounting software, even though customers reach them under your brand. CISA's template sends weaknesses in vendors' systems to the vendor, under the vendor's own disclosure policy, and the template here does the same.
Response targets are numbers you set, and a target you meet does more for trust than a quicker one you miss. CERT/CC's guide says an acknowledgement timeframe of 24 to 48 hours is common for vendors and coordinators. CISA's template, written for federal agencies, uses three business days in its example wording for acknowledging reports from people who share contact details. Pick figures that match who actually watches the inbox, including over holidays, and name a backup.
A first reply is not the end of it. FIRST's guidelines ask every party to acknowledge receipt of each communication and ask vendors to keep the finder updated on status, so the template sets a regular update interval as well as a first-reply target.
A researcher may want to write up what they found once it is fixed. Decide whether you will agree a date for each case or publish a default window. CERT/CC's guide says 45 to 90 days seems to be the normal range for public disclosures, and recommends treating a policy's stated timeframe as the starting point of a negotiation rather than a hard deadline.
Its guide also lists reasons a date moves: evidence of active exploitation, or a vendor that stops responding, can bring it forward, and a vendor still making progress in good faith can be a reason to hold it back. The template offers both an agreed date and a default window; keep the one that suits you.
A researcher looking at a flaw in a customer portal may see a customer's details before they realise what they are looking at. The policy tells them to stop, keep nothing and tell you. Inside the business, the same moment can start your data breach response.
Under the Notifiable Data Breaches scheme, an organisation covered by the Privacy Act 1988 must notify affected individuals and the OAIC when a data breach is likely to result in serious harm to a person whose personal information is involved.
Whether a given report reaches that point is an assessment for your privacy officer and lawyer, so the template keeps that step in an internal note rather than in the published text.
The policy speaks for the business, so someone with authority over each system in scope approves it, along with the people named as contacts. If a developer, agency or host runs part of the stack, agree with them how a report reaches them and how quickly they act on it, because the targets you publish depend on their response as much as on yours.
What a safe-harbour statement can and cannot do
Safe harbour is the part of a policy that says how the organisation will treat a researcher who follows it.
In this template it is a statement of the publishing organisation's own intent: if a researcher makes a good-faith effort to follow the policy, the organisation will not take legal action against them over that research or ask anyone else to, and if someone else does, it will make known that the research followed the policy.
CISA's template carries a similar commitment, and CERT/CC's guide says that if an organisation chooses to explicitly disavow legal retribution against reporters who otherwise follow its policy, the policy should lay that out clearly.
The statement has limits, and the template spells them out. It covers decisions the organisation itself controls. It does not bind police, regulators, suppliers or anyone else, and it does not extend to systems the organisation has left out of scope. disclose.io's policy terms carry the same caveat: their safe harbour applies only to legal claims under the control of the organisation, and the policy does not bind independent third parties.
Some published wording goes further. CISA's template says the agency will consider good-faith research authorised, and disclose.io's terms describe research under the policy as authorised concerning applicable anti-hacking laws and as lawful. The template here leaves those words out. Whether to add a statement of permission or authorisation, and in what words, is a decision for your solicitor, because it describes the research in legal terms instead of stating what your business will do. disclose.io itself says its language is not legal advice, and lists an Australian variant of its terms that is marked as a draft.
For Australian readers, Part 10.7 of the Criminal Code, in the Schedule to the Criminal Code Act 1995 (Cth), is headed Computer offences. Section 478.1, for example, makes it an offence to cause unauthorised access to, or modification of, restricted data where the person intends to cause it and knows it is unauthorised, and section 476.4 says the Part is not intended to exclude or limit the operation of other Commonwealth, state or territory laws, subject to sections 476.6 and 476.7.
This page does not say how those provisions apply to any researcher, report or policy. That is a question for a lawyer.
FIRST's guidelines advise every party to avoid escalation, including legal action, as far as possible, and say that legal or other coercive pressure, actual or perceived, often creates a chilling effect on desired security research.
Publishing security.txt alongside the policy
RFC 9116, an informational RFC the IETF published in April 2022, defines security.txt: a plain-text file that tells researchers how to reach you and where your policy lives.
OWASP's cheat sheet lists it among the places a researcher can look for contact details when there is no clear disclosure policy, and among the ways an organisation can publish its own. In a security.txt policy setup, the file says where to report and its Policy field links to the page that says how.
The RFC describes the file as a complement to an organisation's other public material on its disclosure practices, not a replacement for it.
The checklist in the template covers each step. These are the rules to check before the file goes live:
- Location. For a website, the file goes under the /.well-known/ path, for example https://example.com/.well-known/security.txt. For legacy compatibility the RFC allows for a file at the top-level /security.txt path, or a redirect from there to the /.well-known/ file, and where a file exists in both places the RFC says the one under /.well-known/ must be used.
- Transport and type. The file is fetched over HTTPS and served with a Content-Type of text/plain and the charset set to utf-8.
- Two required fields. Contact must always be present: email addresses use mailto:, phone numbers use tel:, a web address must begin with https://, and several Contact lines should be listed in order of preference. Expires must also be present, exactly once, as a date and time, and the RFC recommends a value less than a year into the future.
- Everything else is optional. Policy links to the disclosure policy, and Encryption points to where a key can be retrieved rather than holding the key itself. Acknowledgments, Canonical, Preferred-Languages and Hiring are also defined.
- One file per host. A security.txt file applies only to the domain or IP address it was retrieved from, not to its subdomains or parent domains, so a customer portal on its own subdomain needs its own file if it is to carry one.
- Stale information. The RFC warns that out-of-date contacts can send reports to the wrong place, and says not having a security.txt file may be preferable to having stale information in it. Put the Expires date in a shared calendar with a reminder well before it.
- No implied permission. Section 5.5 says researchers should not assume that the presence or absence of the file grants or denies permission to test. It says any such permission may be indicated in the organisation's vulnerability disclosure policy or in a new field.
- A target for attackers too. An attacker who has compromised a website can change the file or redirect it. The RFC's answer is the Canonical field, an OpenPGP signature on the file and regular monitoring of the file and the resources it points to.
- More automated reports. The RFC notes that a published file makes reports sent in an automated fashion, or from automated scans without human analysis, more likely, and that organisations need to weigh that extra triage against the benefits.
How a report moves from inbox to close
CERT/CC's guide breaks coordinated disclosure into phases adapted from ISO/IEC 30111: discovery, reporting, validation and prioritisation, remediation, public awareness and deployment.
For a business receiving a report about its own website or application, those phases become a short run of tasks, and the worked example in the template follows one report through them.
- Log it. Give the report a reference, record when it arrived, and acknowledge it within your published target, even if the only news is that it has arrived.
- Reproduce it safely. Confirm the issue with your own test accounts and test records, on a staging copy where you have one, rather than by repeating the researcher's steps against real customer data.
- Rate it and give it an owner. Decide how serious it is for your business and who will fix it, and tell the researcher the outcome.
- Contain, then fix. A temporary measure, such as switching off one feature, can come before the real fix. Check whether the same flaw appears in other parts of the application built the same way.
- Read the logs. Look for signs that anyone other than the researcher used the flaw, as far back as your logs go, and record that limit.
- Close the loop. Tell the researcher it is fixed, agree any credit and disclosure date, and record the cause and the test that now covers it.
What a useful report contains
CERT/CC's guide asks reporters for the affected versions, how the issue was found, reproduction steps or proof-of-concept code, the impact, any suggested fix and any time constraints, such as a planned publication date. OWASP's cheat sheet adds HTTP requests and responses, screenshots and other supporting evidence, with personal data redacted before the report is sent.
Section 5 of the template turns those into a short list a researcher can follow without guessing what you need.
Start from good faith
One of the principles CERT/CC lists is to presume benevolence. It has found that it is usually best to assume that anyone who has taken the time and effort to report an issue is likely benevolent and sincerely wants to reduce the risk. A reply written from that assumption keeps the researcher talking to you rather than about you.
When the weakness belongs to a supplier
Some reports concern software you run but did not write: a WordPress plugin, a JavaScript library, a hosted form or the platform your site sits on.
CISA's template sends these to the supplier under the supplier's own disclosure policy, and the template here does the same, with an offer to pass the report on if the researcher prefers.
FIRST's guidelines describe a coordinator as particularly helpful when multiple vendors are involved or a vendor is difficult to contact, and OWASP's cheat sheet notes that a national or sector-based CERT may be able to assist when an organisation cannot be contacted directly.
In Australia, the Submit a report page of ASD's ACSC includes an option to report a cyber security vulnerability. Ask the researcher before you pass on anything that identifies them.
Before the policy goes live
A policy is a public statement of how your business will behave, so test the arrangements behind it before you publish.
- Send a test report to every address in the policy and in security.txt, from outside the business, and confirm who receives it and how quickly.
- Brief the people who watch the general inbox, contact form, web chat and phones on how to recognise a security report and who to pass it to.
- Agree the response chain with your developer, agency and host, so the targets you publish are ones they can meet.
- Prepare test accounts and a staging copy, so a report can be reproduced without touching real customer records.
- Have the scope approved by someone with authority over each system, and the safe-harbour and privacy sections reviewed by your solicitor.
- Set review dates for the policy and for the Expires field, and put both in a shared calendar.
Testing the systems you are about to put in scope
If the systems you are about to list have never been tested, it is worth testing them first, so the issues a scoped test would find surface on your timetable rather than in a stranger's email.
Clever Ops tests what customers reach through a browser or an API: websites, WordPress sites and WooCommerce stores, custom web applications and their APIs, apps made with AI builders, and AI chatbots and agents.
If a report suggests someone has already got in, call an incident response specialist first. Clever Ops does not provide emergency incident response, and an audit can follow once the site is clean.
Frequently asked questions
If you run a website, customer portal, online store or API, people can find weaknesses in it and will sometimes tell you. A policy decides where that message goes and what happens next. This page does not assess whether a contract, customer or sector rule in your situation asks for one. Check those documents, and ask your solicitor if the wording is unclear.
The clause is your organisation's statement of what it will do: not take or recommend legal action over research that follows the policy, and make known that the research followed it if someone else acts. It does not bind police, regulators or anyone else, and this page does not say how any law applies to a researcher. A researcher who wants certainty should get their own legal advice.
The template offers no payment and says so. Paying for reports turns a policy into a bug bounty, which OWASP recommends only for organisations that already have a mature disclosure process. OWASP lists non-cash thanks, such as credit in a hall of fame or similar acknowledgement, and RFC 9116 defines an Acknowledgments field that links to a page where researchers are recognised for their reports.
Hold off and escalate it inside the business. OWASP's cheat sheet says such demands may be legitimate but in many cases are scams, and suggests asking the person to disclose through a mediated bug bounty platform. Its advice to researchers is that demanding payment before revealing the details may constitute extortion. Keep every message, check your systems for the issue described, and treat any sign of real access as an incident.
CERT/CC's guide says there is no single right answer to how long is long enough to respond to a vulnerability, and the template asks you to set your own targets. On first replies, the same guide says acknowledgement within 24 to 48 hours is common for vendors and coordinators, and CISA's template for US federal agencies uses three business days in its example wording. Choose targets your team can meet consistently, including over holidays, and tell the researcher early if a fix will take longer.
A Contact line and an Expires line, served at /.well-known/security.txt over HTTPS. RFC 9116 makes those the two fields that must always be present, and recommends an Expires value less than a year into the future. Adding a Policy line that links to your disclosure policy tells researchers what you allow.
No. A disclosure policy is public and written for people outside the business who find a weakness. A security policy is an internal document that sets how your own staff and suppliers handle access, changes, logging and testing. The security policy template covers the internal document, and it can point to the disclosure policy for reports from outside.
Scope the right test for your application
Tell us what you have built, who uses it and when it needs to launch. You will get a plain-English recommendation: a website audit, a web application test or a referral to a specialist.