Read a sample: 1. The Authorized Tester's Working Boundaries
A web request can look deceptively simple: a method, a path, a few headers, and perhaps a body. Yet capturing or changing that request can expose private information, alter application state, or send traffic to a system that someone else operates. Before a tester interprets the first exchange, the tester needs to know why the work is authorized, which systems are included, and what must happen to the evidence afterward.
These boundaries are not administrative details separate from technical work. They determine which observations are valid, which experiments are permitted, and whether another person can trust the resulting report. A carefully documented response from the wrong host is not a useful finding. Neither is a convincing demonstration that exceeds the agreed limits.
This chapter establishes a working approach for the rest of the book. You will define permission and scope, distinguish scope from rules of engagement, prepare a small local practice environment, and create an observation notebook. The examples use fictional applications and accounts. Adapt the process to real engagements only under the relevant owner’s explicit authorization.
Authorization is an agreement that permits specified testing activity against specified assets. It should come from a person or organization with the authority to grant it, and it should be documented in a form appropriate to the engagement. A public website, a login page, or an exposed application programming interface is not, by itself, permission to test.
The practical question is not merely, “May I look at this application?” It is, “Who has authorized which people to perform which actions, against which assets, during what period, under which constraints?” A useful written authorization identifies the parties involved and connects the testing activity to an agreed purpose. It also provides a way to resolve uncertainty before the tester acts.
For example, an application owner might authorize a tester to inspect a staging version of a customer portal using two supplied test accounts. That authorization does not automatically include the production portal, the organization’s other domains, its cloud provider’s management interfaces, or a third-party identity service used by the portal. Even closely related systems can have different owners and different conditions for testing.
Authorization can also be conditional. A system owner may permit traffic capture and manual input changes while excluding automated scanning, password guessing, or testing that alters data. The tester should work within those conditions rather than treating them as suggestions. If the proposed exercise requires an action that is not clearly permitted, pause and obtain clarification before proceeding.
In a personal lab, the situation is simpler but the principle is the same. Use an application and environment you control, or one expressly provided for practice, and ensure that dependent services and data are also appropriate to use. Installing a local application does not make every external service it contacts part of the lab.
Scope describes the assets and activities included in a test. A scope statement should be specific enough that two people can independently decide whether a proposed request belongs in the engagement.
Consider this fictional scope:
> Manual observation and limited input testing are authorized for the staging portal at https://staging.example.test during the agreed test window. Use only the two supplied fictional accounts. Do not test production systems, third-party identity services, payment processing, or availability limits. Report unexpected access to another user’s data without collecting additional records.
This statement gives the tester a starting point, but a real engagement may need more detail. Does the staging hostname resolve only to infrastructure owned by the authorizing organization? Are particular API hosts included? Are uploaded files allowed? May the tester modify records created by the test accounts? What should happen if the application redirects to an external service?
Clarify these questions before testing rather than relying on a domain name or interface label. A browser workflow can cross several hosts without making the transition obvious. A page within scope may load scripts, send analytics, or authenticate users through services outside it. The fact that a request originates from an in-scope page does not automatically bring its destination into scope.
A practical scope record can include:
- Assets: Exact hostnames, applications, environments, and any relevant network ranges.
- Accounts and data: Approved identities, roles, test records, and restrictions on personal or production data.
- Activities: Permitted observation, input changes, automation, and state-changing actions.
- Exclusions: Systems, endpoints, services, techniques, and data types that must not be tested.
- Timing: The authorized window, time zone, and any operational restrictions.
- Contacts: The person who can clarify scope, an operational contact, and an escalation route.
Keep both inclusions and exclusions visible. If a hostname, account, or action is absent from the record, do not assume it is included. Ask the engagement contact and record the answer.
Scope answers where and broadly what. Rules of engagement explain how the work will be performed. They turn permission into operational decisions that a tester can follow when a request produces an unexpected result.
For request analysis, useful rules address whether the tester may intercept traffic, repeat requests, change parameters, upload content, create records, or use automation. They also specify limits on request volume and concurrency where appropriate. A rule such as “avoid disrupting service” is a sound objective, but the people responsible for the application should help translate it into workable limits and a stop procedure.
A rules-of-engagement discussion should also cover:
- How to recognize and report…