Security

Pentest Rules of Engagement: What to Agree On Before Testing Starts

OraSecOctober 20, 20269 min read
Pentest Rules of Engagement

Pentest rules of engagement are the written agreement that says what testers may do, to which systems, when, and how. They turn "please test our security" into clear permission, covering scope, allowed techniques, emergency contacts, and data handling. Without them, a test can hit a system the client doesn't own or run an exploit at the worst possible time. Agreeing them is one of the first things to settle when you prepare your organization for a pentest.

What are rules of engagement in penetration testing?

The signed rules a pentest runs under: what testers may test, which techniques they may use, and when.

What should be included in pentest rules of engagement?

Scope and exclusions, written authorization, permitted activities, testing windows, accounts, data handling, contacts, emergency stop, and reporting.

What is the difference between pentest scope and rules of engagement?

Scope defines what gets tested. Rules of engagement define how testing is done, when, and under what limits.

Who signs the rules of engagement for a penetration test?

The system owner or an authorized delegate signs for the client, and the testing firm's lead signs for the testers.

What happens if a tester finds an out-of-scope vulnerability?

The tester stops, doesn't exploit it, and reports it. Testing continues only after written approval to add it.

Key Takeaways

  • Rules of engagement are the testers' written permission. If an action isn't covered, it isn't authorized.
  • Scope says what gets tested. Rules of engagement say how. Agree and sign both before testing starts.
  • The signer must own or control the systems. Cloud and hosting providers have their own rules.
  • Name contacts on both sides and agree an emergency stop process before testing begins.
  • Out-of-scope findings get reported, not exploited. Scope changes get approved in writing.
  • Vague wording can create disputes even when the testing itself is done properly.

What Are Pentest Rules of Engagement?

Pentest rules of engagement (RoE) are the agreed terms a penetration test runs under. They sit alongside the contract and answer the questions testers face once they start: Can we exploit this? Can we phish staff? Who do we call at 2 a.m. if a server goes down?

NIST SP 800-115 includes a rules of engagement template in Appendix B. A penetration test needs valid authorization from someone entitled to approve it and must comply with applicable laws and contracts. In the UK, that's the line between a pentest and an offense under the Computer Misuse Act 1990. The rules of engagement are where that permission is written down in enough detail to be useful. They're agreed during scoping, the first stage of the penetration testing methodology, and should be specific enough that a tester never has to guess.

Why Pentest Rules of Engagement Matter

A well-known example isn't technical. In September 2019, two Coalfire testers were arrested at the Dallas County Courthouse in Iowa during a physical security test for the state's court administration. The sheriff argued the state had no authority to approve after-hours entry to a county building. The charges, burglary later reduced to trespass, were dropped in January 2020, and the testers reached a $600,000 settlement with the county and its former sheriff in January 2026. They had a signed authorization letter. What the paperwork didn't settle was who owned the building, whether that owner had agreed, and whether after-hours entry was allowed.

Other failures are less dramatic but follow a similar pattern:

  • Legal exposure. Testing a system the client doesn't control can break the law.
  • Business disruption. An aggressive scan during a payroll run can cause an outage.
  • False alarms. An uninformed SOC may escalate to incident response or law enforcement.
  • Wasted findings. Stopping at "vulnerability identified" hides how far an attacker could get.

Pentest Rules of Engagement vs Scope of Work

The two are often merged into one document, but the scope of work is the commercial agreement and the rules of engagement are the operating rules.

Area

Rules of Engagement

Scope of Work

Purpose

How testing is done safely and legally

What the client buys and receives

Scope and boundaries

Limits: excluded systems, forbidden techniques, data limits

Targets: applications, IP ranges, environments, roles

Testing activities

Allowed activities, such as exploitation, phishing, or lateral movement

Test type, such as external, web app, or cloud, and objectives

Responsibilities

Contacts, escalation paths, and who can stop the test

Project owner and commercial contacts

Authorization

Signed permission from the system owner and third parties

The commercial agreement to deliver the service

Deliverables

How findings, including critical issues, are shared mid-test

Final report, debrief, retest, and timelines

Changes and approvals

Written approval before adding any target or technique

Changes to cost, effort, and timeline via change request

In practice, commercial terms, deliverables, and project timelines belong in the scope of work. Rules of engagement define what testers are authorized to do, when testing is permitted, and which restrictions apply.

What to Agree On Before Testing Starts

These eight areas are worth settling before testing begins. Leaving one out can cause problems mid-test.

1. Define In-Scope and Out-of-Scope Assets

List every target precisely: hostnames, URLs, IP ranges, cloud account IDs, API base paths, app versions, and environments. "Our website" isn't a scope; "app.example.com, api.example.com/v2, and 203.0.113.0/26 in production" is. List exclusions just as clearly, such as shared infrastructure or third-party payment pages. Agree what happens with assets nobody listed. In one case, a fintech's test server holding real customer data was built outside the standard deployment process and never added to the inventory; OraSec's threat intelligence team traced it only after that data was offered on an underground forum. The rules should say whether testers may report or verify such assets.

2. Confirm Written Authorization and Third-Party Permissions

Authorization must come in writing from someone who owns or controls the systems, naming the targets and dates. Then check every third party. Standard testing of your own AWS and Azure resources generally needs no provider pre-approval, but service-specific restrictions apply: AWS bans denial of service and some DNS and takeover attacks, and requires a Simulated Events form for activities such as simulated phishing or red team command and control. SaaS platforms, managed service providers, and building owners may need to agree too, as our guide to cloud penetration testing rules covers in more detail.

3. Agree Testing Objectives and Permitted Activities

Settle the key questions: Is exploitation allowed, or only identification? Can testers move laterally and escalate privileges? Are password spraying and offline cracking allowed, and against which accounts? Are phishing, vishing, wireless attacks, or physical entry in scope? Can testers create accounts or install persistence for proof? Is denial of service allowed (usually not)? These choices change what a test can prove. In one Active Directory takeover, an evil twin Wi-Fi access point captured an employee's WPA2-Enterprise login, the password was cracked offline, and an AD CS certificate template misconfiguration led to Domain Admin. A test that excluded wireless attacks would not have shown that chain. Tie each activity to an objective, such as "reach Domain Admin."

4. Establish Testing Dates, Windows, and Restrictions

Agree start and end dates, daily testing hours, and time zone. Add blackout periods, such as month-end close, release freezes, or peak sales. If some tests are only safe after hours, confirm someone can respond. Record stability limits, such as scan rates.

5. Confirm Test Accounts and Access Requirements

Agree the accounts, roles, and tenants testers will use, and when they're disabled. Access control testing often compares two users in the same role and users across roles, which is how flaws like broken object-level authorization are found, so ask for two accounts per role. Also confirm source IPs, VPN access, and MFA arrangements.

6. Define Data Handling and Confidentiality

Decide how much sensitive data testers may access to prove a finding, such as one redacted record rather than a full export. Agree where evidence is stored, who can see it, and when it's deleted, including recovered credentials and hashes. The report itself is sensitive too.

7. Establish Communication and Emergency Stop Procedures

Name a primary and backup contact on each side and agree how critical findings are reported mid-test. Most importantly, agree the emergency stop: who can call it and what happens next. If a system slows down, a tester finds signs of a real attacker, or the SOC raises an incident, testing pauses until both sides agree to continue. Decide whether the SOC will be told.

8. Agree Reporting, Remediation, and Retesting

Agree the report format, severity ratings, recipients, and delivery date. A penetration testing report example shows what a finding should contain: evidence, impact, severity, and a fix. Then agree the debrief, remediation window, and retest.

Helpful for you: What Happens After a Penetration Test Ends

How to Handle Scope Changes During a Pentest

Scope changes are normal. Testers may find a forgotten subdomain, a reachable internal network, or a staging server that mirrors production, a common target because attackers love non-production environments. A sound process looks like this:

  1. Stop. Don't test the new asset further.
  2. Record. Note what was found and how.
  3. Notify. Tell the named contact straight away.
  4. Confirm ownership. The client checks who owns it.
  5. Approve in writing. An authorized owner or delegate approves the added asset, with any required third-party permission. A verbal "go ahead" isn't enough.
  6. Adjust. Update time and cost if needed.

If the client removes a system mid-test, record it so the report shows what was and wasn't tested.

Pentest Rules of Engagement Checklist

  • In-scope assets listed precisely.
  • Out-of-scope assets and techniques listed.
  • Ownership of every in-scope asset confirmed.
  • Written authorization signed by someone with authority over the systems.
  • Third-party and cloud terms checked.
  • Test objectives agreed, such as "access customer data."
  • Permitted activities agreed.
  • Forbidden activities agreed, such as denial of service.
  • Dates, testing hours, time zone, and blackout periods set.
  • Rate limits and stability restrictions recorded.
  • Test accounts provided where required, ideally two per role for access control testing.
  • Access tested before day one.
  • Data handling rules agreed.
  • Primary and backup contacts named on both sides.
  • Emergency stop process agreed, including who can call it.
  • Process for critical findings during the test agreed.
  • Decision made on whether the SOC is told.
  • Process for out-of-scope findings and scope changes agreed.
  • Report format and delivery date agreed.
  • Debrief, remediation window, and retest agreed.
  • Final document signed by both sides and shared with every tester.

Common Pentest Rules of Engagement Mistakes

  • Vague targets. "The website" leaves testers guessing.
  • Authorization from the wrong person. A developer can't authorize testing of systems they don't control.
  • Forgetting third parties. Providers and landlords have their own rules.
  • Silence on techniques. Unmentioned techniques can't be assumed allowed.
  • No emergency contact who answers. A contact on leave during the test is no contact at all.
  • No stop process. Nobody knows who can halt testing or how.
  • Ignoring the SOC. It can trigger wasted incident response or hide a real attack.
  • Verbal scope changes. "Just have a look at that one too" on a call isn't written approval.
  • No data handling rules. Customer data can end up in emails.
  • Copying last year's document. It can point to old systems and contacts.

Pentest Rules of Engagement Best Practices

  • Start early. Agree the rules during scoping, since third-party approvals take time.
  • Be specific. Use exact hostnames, IPs, account IDs, and dates.
  • Write down what's forbidden, not just what's allowed. Clear exclusions help prevent serious mistakes.
  • Tie activities to objectives. Testers aim better when they know what you need to learn.
  • Run a kickoff call. Walk through the document, check access works, and confirm contacts.
  • Keep one version. Every tester uses the latest signed copy.
  • Give testers proof of authorization. On-site testers should carry the signed letter and a 24-hour contact.
  • Review after the test. Fix anything unclear in the template before the next engagement.

How OraSec Can Help

OraSec agrees the rules of engagement with you during scoping for every engagement: targets, exclusions, permitted techniques, testing windows, contacts, data handling, and emergency stop. We check cloud and third-party terms, confirm access before day one, and tell you straight away about critical issues or assets outside the agreed scope. The result is a test that goes as deep as you need without surprises for your operations or suppliers. The team's own research has produced published CVEs (CVE-2025-50847 through CVE-2025-50850). If you're planning a pentest, talk to our team.

Conclusion

Rules of engagement help ensure a penetration test is properly authorized, conducted safely, and produces useful results. They tell testers how far they can go and tell the client what will happen to its systems. Agree targets, authorization, techniques, dates, and data rules, and make sure someone answers the phone if testing must stop. Then the test can focus on finding the weaknesses an attacker would use.

Frequently Asked Questions

How far in advance should rules of engagement be agreed?

Ideally one to two weeks before testing, or earlier with third parties. AWS requires Simulated Events requests at least two weeks before the start date.

Do we need to tell our cloud provider before a pentest?

Usually not for standard testing of your own AWS and Azure resources, but service-specific restrictions apply, and some activities need approval.

Should our SOC know the pentest is happening?

Usually the security lead should. To test detection, keep the wider team unaware, but someone senior must be able to confirm authorization.

What happens if a pentest causes an outage?

Testers stop, call the named contact, and follow the emergency stop process. Testing resumes only when both sides agree it's safe.

Are rules of engagement different for a red team engagement?

Often, yes. Red teaming can include social engineering, physical access, and covert command and control, so it needs tighter authorization and deconfliction rules.

Does a retest need new rules of engagement?

Not always. If targets and contacts are unchanged, extend the original in writing. If systems, owners, or providers change, update and re-sign it.

Explore related services

Need hands-on help? Our security testing services put this research into practice.

Binance hack 2026

Binance Hack: What We Know About the $6.3 Billion Hot Wallet Theft

On October 29, 2026, attackers drained Binance hot wallets across multiple blockchains in what blockchain analytics firms call the largest crypto theft on record. Binance puts the impact at about $6.3 billion, with roughly $3 billion frozen, and says early findings point to a possibly compromised engineer's laptop, with no evidence of leaked private keys so far. Here's what happened, what's still unconfirmed, and what to check in your own environment. What happened at Binance? Attackers drain

·6 min read
Oracle E-Business Suite RCE (CVE-2026-101767): What to Do Now

Oracle E-Business Suite RCE (CVE-2026-101767): What to Do Now

CVE-2026-101767 is a critical flaw in Oracle E-Business Suite (EBS), rated CVSS 9.8. It lets an unauthenticated attacker send a crafted web request that triggers SQL injection and can be chained into full remote code execution. Oracle patched it in its October 20, 2026 Critical Patch Update, and attackers were already probing for it a week later. If you run EBS, this one needs attention now. What is CVE-2026-101767? An unauthenticated SQL injection in Oracle E-Business Suite that can be chain

·5 min read
Black October 2026: Every Major Vulnerability and Breach

Black October 2026: Every Major Vulnerability and Breach

October 2026 brought so many critical flaws and breaches at once that the security community started calling it "Black October." Two unauthenticated remote code execution bugs in PHP core, a CVSS 10.0 flaw in GitLab reaching back twelve years, an exploited SQL injection in Oracle E-Business Suite, Microsoft's largest-ever Patch Tuesday, and a ~20 million patient breach at Oracle Health all landed in the same few weeks. This roundup covers what broke, who's affected, and what to fix first. What

·6 min read