Scope, Contracts, and Connected Assets

A client has authorized a penetration test. You have a signed confidentiality agreement, a contract covering the business relationship, a statement describing the project, and a list containing a domain and an Internet Protocol address range. During reconnaissance, you also disc…

A client has authorized a penetration test. You have a signed confidentiality agreement, a contract covering the business relationship, a statement describing the project, and a list containing a domain and an Internet Protocol address range. During reconnaissance, you also discover a related cloud service and a mobile application. Can you test everything that appears connected to the client?

Not from those facts alone. Before choosing a technical action, you need to identify which document governs which part of the engagement, translate the target language into a precise inventory, and determine what kind of assessment was actually authorized. A technical relationship between two systems isn’t the same thing as contractual permission to test both systems.

For PenTest+ PT0-003, this subject belongs to pre-engagement activity. The relevant terms include a nondisclosure agreement, or NDA, a master services agreement, or MSA, a statement of work, and terms of service. These documents can interact, but they don’t perform identical jobs. Their exact legal effect depends on their wording and the applicable circumstances, so the operational distinctions here shouldn’t be treated as legal advice or as a substitute for reading the signed documents.

An NDA addresses confidential information. In a penetration-testing engagement, that may include information exchanged before testing, technical data collected during the assessment, or details contained in the eventual report. Its presence tells you that confidentiality obligations exist. It does not, by itself, identify the systems you may test or authorize a particular technique. A tester holding a signed NDA can still be completely unauthorized to scan a system.

An MSA generally establishes terms for the broader business relationship. It can provide a common foundation across multiple pieces of work, rather than renegotiating every relationship term for every project. That makes it important, but it doesn’t mean every system owned by the client is included in every engagement. If the MSA establishes a relationship between the testing company and the client while the statement of work identifies one particular assessment, the engagement-specific boundaries still matter.

The statement of work defines the work and deliverables for the engagement. This is where you’d expect to find project-specific information such as what work is being requested and what result is to be delivered. Depending on how the documents are organized, some operational details may appear elsewhere, including in the rules of engagement. You have to consider the documents together rather than assume that one familiar heading contains every controlling detail.

Terms of service govern use of a service or platform. They matter when testing touches infrastructure or functionality operated by another provider. A client’s authorization doesn’t erase a provider’s conditions, and a provider allowing certain security testing doesn’t give you permission to test another customer’s resources. In a cloud engagement, both the client’s authority over the named resources and the provider’s applicable testing conditions can affect the decision. The actual platform, resource, and proposed activity have to be checked rather than inferred from the word cloud.

A useful way to separate these documents is to ask what question each one helps answer. How must sensitive information be handled? Look to the NDA. What terms govern the continuing commercial relationship?

That’s the role of the MSA. What work and deliverables belong to this engagement? Examine the statement of work.

What conditions apply to using a provider’s service? Review the terms of service. None of those questions eliminates the need for explicit authorization, scope, rules of engagement, timing, and escalation paths.

When several documents are present, one may sound broader than another. An MSA may describe security services in general, while the statement of work names only this project and its deliverables. That broader language doesn’t silently enlarge the target list. Read the documents together and identify the engagement-specific authority before acting.

The same reasoning applies to confidentiality. An NDA may require you to protect information you’ve already received or observed, but that duty doesn’t become permission to collect more of it. If the documents appear inconsistent, preserve the ambiguity and use the agreed clarification path rather than choosing whichever sentence permits the most testing.

Now consider the target list. PT0-003 identifies CIDR ranges, domains, individual IP addresses, and URLs as forms of target selection. Those forms aren’t interchangeable. Each identifies something at a different level of specificity, and each needs to be interpreted in the context of the written scope.

CIDR means Classless Inter-Domain Routing. CIDR notation expresses a range of IP addresses by combining an address with a prefix length. If the scope names one small authorized network block using CIDR notation, the prefix identifies its address boundary. Your working inventory should record the exact range as written, along with any exclusions or qualifications. It shouldn’t quietly grow to include adjacent addresses just because they look related or appear to be operated by the same organization.

An individual IP address is narrower. It identifies one network address, but it doesn’t automatically tell you which applications, virtual hosts, or services the client intended to include. Conversely, a domain name identifies a name rather than every possible address, hostname, or external service associated with the organization. The scope might name a domain broadly, or it might name one exact host under that domain. Those are materially different descriptions, and ambiguous wording needs clarification before stronger testing begins.

A URL can be more specific because it may identify a particular scheme, hostname, and path. If the target list names one web address under a particular host, don’t assume that every sibling hostname, unrelated path, administrative interface, or back-end service is included. The written scope may authorize broader coverage, but the URL alone shouldn’t be stretched beyond what it actually identifies.

Translate all of this into an explicit target inventory before testing. For each entry, record the literal identifier, the assessment type connected to it, any exclusions, and any uncertainty requiring clarification. A domain entry remains a domain entry.

An IP range remains the defined range. A specific URL remains specific unless the governing documents clearly establish a broader boundary. This inventory gives technical staff and engagement contacts something concrete to verify before activity begins.

That inventory also exposes mismatches before they become testing mistakes. Suppose one row contains the authorized network range, another contains the named example domain, and a third contains a specific URL. The first row supports decisions about addresses in the defined block. The second records the named domain boundary as the scope actually states it.

The third preserves the narrower web location. If reconnaissance links all three, you can record the relationship without merging them into one unlimited target. Then attach the assessment type and any exclusion to the relevant entry. A reviewer can see whether a proposed network check, web check, or request for clarification follows from the written identifiers instead of from an assumption about ownership.

Suppose the statement of work names an example domain and one small authorized network block. Open-source reconnaissance then reveals a hostname that appears to belong to a software-as-a-service platform used by the client. The hostname looks related to the client’s brand, and a record may even associate it with infrastructure already observed during reconnaissance. What can you conclude?

You can document that the hostname was discovered and that it appears related. You can’t conclude that it has joined the authorized target list. It could be operated by a third party, shared with other customers, governed by separate provider conditions, or simply absent from the agreed scope. Those possibilities don’t prove that testing is forbidden forever. They show why the current evidence is insufficient to authorize it now.

The defensible next action is bounded. Compare the discovery with the signed scope and target inventory. Check whether the wording explicitly includes that hostname, a broader domain boundary, the relevant application, or the provider-hosted resource.

Check the timing and operational constraints as well. If the documents remain unclear, use the established escalation or clarification path. Don’t silently convert technical discovery into contractual permission.

Even name resolution doesn’t settle the issue. Imagine that a newly discovered hostname resolves to an IP address inside the authorized network range. That result establishes a technical relationship at the time of the lookup.

It doesn’t establish the client’s legal authority over the hostname, prove that every service reached through the address is in scope, or replace the written agreement. Shared hosting, changing records, and incomplete documentation are all reasons to distinguish what you observed from what you’re authorized to test. The controlling point is simpler: scope comes from the engagement documents, not from DNS resolution alone.

Your validation should match the uncertainty. If the question is whether an identifier appears in the approved inventory, read and reconcile the governing documents first. If the question is whether an observed hostname corresponds to a listed target, use the least disruptive evidence that can answer that question within the rules. A more aggressive technique isn’t automatically more convincing. When a low-impact method can establish the needed fact, use it before considering a higher-impact action.

Conflicting evidence should slow the conclusion, not expand the test. A stale reconnaissance result might point to an address that has changed. A tool might normalize a domain in a way that hides an important subdomain distinction.

A tester might copy the wrong prefix length or assume that a broad company name means a broad technical scope. Recheck the target, source data, configuration, prerequisites, and freshness of the observation. Then record what is confirmed, what is contradicted, and what remains unresolved.

This same discipline applies to assessment types. PT0-003 distinguishes web, network, mobile, cloud, application programming interface, application, and wireless assessments. These categories can overlap technically, but authorization for one doesn’t automatically authorize all the others.

A network assessment focuses on the network targets and exposed services identified by the engagement. A web assessment focuses on web-facing functionality within the named boundary. A broader application assessment may concern software behavior beyond a single web interface. An application programming interface, or API, assessment focuses on the authorized interfaces through which software components exchange requests and responses. The presence of an API behind an authorized website doesn’t establish that every endpoint, environment, or connected integration is included.

A mobile assessment may involve a mobile application and the systems it communicates with, but naming the mobile application doesn’t automatically authorize testing every back-end service it can reach. The package, test accounts, supported environment, API endpoints, and related infrastructure need their own clear boundaries. The same principle works in reverse. Authorization for a back-end web service doesn’t necessarily include analysis of a separately distributed mobile application.

Imagine the authorized web assessment shows that the named site sends requests to an API used by the mobile application as well. That observation is useful, but it doesn’t collapse three assessment types into one. You can record the request destination and compare it with the approved target inventory. If the API endpoints aren’t named or otherwise clearly covered, deeper API testing waits for clarification.

If the mobile package isn’t included, the website’s connection to a mobile back end doesn’t authorize analysis of the package either. The decision follows from both dimensions of scope: which target is named and which kind of assessment is approved. Matching only one dimension leaves a gap, even when the systems are technically connected.

A wireless assessment concerns the authorized wireless environment. A company name or office address isn’t enough to identify every nearby network as a target. The target inventory needs to distinguish the client’s approved wireless systems from neighboring, guest, third-party, or otherwise excluded networks. Operational constraints also matter because a technically valid test can still create unacceptable effects if it violates the agreed testing conditions.

A cloud assessment requires particular care about resource ownership and provider boundaries. The word cloud doesn’t identify one uniform permission model. The engagement must name resources the client owns or is authorized to have tested, while applicable provider rules may restrict certain activities or require additional handling. Discovering a client-branded hostname on a cloud platform doesn’t make the entire platform, tenant, or provider infrastructure fair game.

Consider two proposed actions. The first is to begin active testing against the newly discovered software-as-a-service hostname because it shares branding with the client. The second is to document the hostname, compare it with the written inventory, and request clarification before active testing. Based on the available evidence, the second action is stronger. It respects the difference between discovery and authorization while preserving useful reconnaissance information for the client.

Additional information could change that decision. An amended statement of work might explicitly add the hostname. The client might provide written confirmation through the designated authorization path, together with any provider conditions and updated operational boundaries. Until that evidence exists, the technically reachable target remains an observed asset, not an authorized one.

The resulting records matter throughout the engagement. The target inventory guides preparation and prevents different testers from interpreting the same scope in incompatible ways. Notes about unresolved ownership or provider restrictions support escalation. During analysis, those records help separate validated findings from output collected against the wrong target or under an unsupported assumption. In reporting, they let you describe what was tested, what evidence supports the result, and what remained outside the assessment.

If an unexpected service appears on an authorized address, the first reporting question isn’t automatically, “How serious is the vulnerability?” First establish whether the service and assessment type were within the authorized boundary and whether the evidence was gathered under the agreed constraints. An observation made during approved activity can still require clarification before deeper validation. That distinction protects the client, third parties, and the credibility of the final report.

This order matters when the final report is reviewed by someone who wasn’t present during testing. A confirmed finding should rest on evidence gathered from an authorized target with an authorized kind of assessment. An unresolved service can still be documented as an observation, along with the scope question that prevented deeper validation. That wording doesn’t erase useful evidence or overstate it. It tells the client what was seen, why validation didn’t go further, and what authorization would need clarification before any stronger conclusion could be supported.

Contracts, target identifiers, and assessment labels therefore produce practical working decisions. Read the agreement set according to the role each document plays. Convert target language into a reviewable inventory instead of relying on organizational association. Match every proposed action to the authorized assessment type, testing window, provider conditions, and operational limits. Then preserve the evidence behind the decision, including uncertainty and any clarification received, so the next phase begins from a verified boundary rather than an assumption.

A penetration test is authorized by more than a client name or a technical connection. The agreement set tells you how confidentiality, the broader relationship, engagement-specific work, and provider conditions fit together. The target inventory then turns CIDR ranges, domains, IP addresses, and URLs into boundaries that can be checked. Finally, the assessment type tells you what kind of activity those boundaries support.

When reconnaissance reveals something new, preserve the observation without promoting it into scope. Compare it with the written authorization, choose the least disruptive valid check, and escalate uncertainty through the agreed path. That keeps technical evidence useful, the report defensible, and the engagement inside the authority that made the work legitimate.

Before moving on, ask yourself: Can you test everything that appears connected to the client? Use the companion episode reference for exact terminology, named tools or standards, source-derived evidence checkpoints, and review material for Agreements, Target Selection, and Assessment Types. This lesson was produced by Bare Metal Cyber dot com.

Scope, Contracts, and Connected Assets

headphones Listen Anywhere

More Options »
Broadcast by