---

---

Microsoft has established some high enterprise pricing and licensing hurdles going into its FY27. Directions' Advisory Services Director Lane Shelton talks with Mary Jo Foley about the obstacles and potential opportunities.



Download audio: https://www.directionsonmicrosoft.com/wp-content/uploads/2026/07/season5ep12shelton.mp3

---

A Postman collection is a great place to explore an API. But it's a poor place to keep your tests.

Most teams find this out the slow way. Someone exports the collection, converts the requests into test code once, and moves on. Six months later the tests are red, nobody trusts them, and they get skipped in the pipeline. The conversion was never the hard part. Keeping the suite alive is.

This tutorial takes you from a Postman collection to a pytest suite that still passes next quarter. First we'll look at why converted tests rot, then at four principles that keep them alive. The examples stay small, so you can try them on your own collection today.

Table of Contents

• Before You Start

• Why Converted Tests Go Stale

• Principle 1: Keep the Environment Out of the Tests

• Principle 2: Assert on the Contract, Not Just the Status Code

• Principle 3: Make Each Test Stand on its Own

• Principle 4: Put the Suite in Continuous Integration on Day One

• Let a Tool Do the Mechanical Part

• Wrapping Up

Before You Start

To follow along you will need:

• Python 3.10 or newer, with pytest and httpx installed (pip install pytest httpx).

• A Postman collection you want to convert, with its environment (the base URL and token).

• Basic pytest knowledge: how fixtures work and how to run pytest from the command line.

• A GitHub repository if you want to try the continuous integration step. You can skip that part and still follow the rest.

The diagram shows the two parts of the job. On the left, a Postman collection (its requests and environment) is converted into a generated pytest suite, which is the first draft. That conversion is the easy step.

The work is the maintainability layer on the right, which turns that first draft into a suite you can rely on: the environment lives in fixtures instead of being hardcoded, tests assert the response contract rather than just a 200 status, each test is independent, and the suite runs in continuous integration on every push.

Why Converted Tests Go Stale

When you convert Postman requests one to one, you tend to inherit four habits that feel fine on day one and hurt by day thirty:

• The base URL and the token are hardcoded into every test, so moving from staging to production means a find and replace.

• The tests run in a fixed order because request two depends on a value request one set, so a single failure cascades.

• The only assertion is that the status code was 200, which passes even when the response body is wrong.

• Setup is copied into every test, so one change to how you authenticate means editing twenty files.

Every one of these is a maintenance problem, and together they're why the suite gets abandoned. Here's how to avoid each one.

Principle 1: Keep the Environment Out of the Tests

A Postman collection carries its environment in a separate file: base URL, tokens, and other variables. Do the same in pytest. Read those values once, in a fixture, and let every test ask for them.

# conftest.py
import os

import httpx
import pytest


@pytest.fixture(scope="session")
def base_url():
    return os.environ["API_BASE_URL"]


@pytest.fixture(scope="session")
def auth_headers():
    return {"Authorization": f"Bearer {os.environ['API_TOKEN']}"}


@pytest.fixture()
def http():
    with httpx.Client(timeout=10) as client:
        yield client

Now a test never mentions a URL or a token directly:

def test_get_user(base_url, auth_headers, http):
    response = http.get(f"{base_url}/users/1", headers=auth_headers)
    assert response.status_code == 200

Switching from staging to production is now one environment variable, not a search across the whole suite.

Principle 2: Assert on the Contract, Not Just the Status Code

A status of 200 tells you the server answered. It doesn't tell you the answer was correct. The most common reason a broken API ships is that every test only checked the status.

Assert on the shape of the response and the fields your callers depend on.

def test_user_shape(base_url, auth_headers, http):
    response = http.get(f"{base_url}/users/1", headers=auth_headers)

    assert response.status_code == 200
    body = response.json()
    assert set(body) >= {"id", "email", "created_at"}
    assert isinstance(body["id"], int)
    assert "@" in body["email"]

You don't need a strict schema for every endpoint. Even a few checks on the fields that matter will catch a whole class of regressions that a status check waves through.

Principle 3: Make Each Test Stand on its Own

In Postman, it's normal for one request to feed the next. In a test suite, that coupling is a trap: reorder the tests, run one in isolation, or lose the first request, and the rest fall over.

Give each test the state it needs. If a test needs a user, it creates one.

def test_delete_user(base_url, auth_headers, http):
    created = http.post(
        f"{base_url}/users",
        headers=auth_headers,
        json={"email": "temp@example.com"},
    )
    user_id = created.json()["id"]

    response = http.delete(f"{base_url}/users/{user_id}", headers=auth_headers)
    assert response.status_code == 204

Independent tests can run in any order and in parallel, and a failure points at one thing instead of a chain.

Principle 4: Put the Suite in Continuous Integration on Day One

A test suite that only runs on your laptop drifts out of date the moment you stop looking at it. Wire it into your pipeline before you write the second test, so every push has to keep it green.

# .github/workflows/tests.yml
name: API tests
on: [push, pull_request]

jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-python@v5
        with:
          python-version: "3.12"
      - run: pip install -r requirements.txt
      - run: pytest -v
        env:
          API_BASE_URL: ${{ secrets.API_BASE_URL }}
          API_TOKEN: ${{ secrets.API_TOKEN }}

Once this is in place, a test that breaks is a conversation on a pull request, not a surprise in production.

Let a Tool Do the Mechanical Part

Everything above is the part worth your attention. Turning each request into a first draft of a test is mechanical, and mechanical work is worth automating.

I maintain an open-source tool for exactly this step called postman2pytest. It reads a Postman collection and writes a runnable pytest file, so you start from generated tests and spend your time on the maintainability layer rather than on the boilerplate. When the collection changes, you regenerate rather than hand-patching the drift.

You can find it here: https://github.com/golikovichev/postman2pytest

Wrapping Up

Converting a Postman collection into tests is easy. Keeping those tests trustworthy is the real skill, and it comes down to a few habits: keep the environment out of the tests, assert on the contract and not just the status code, make each test independent, and run everything in continuous integration from the start.

Do that, and the suite you generate this week will still be the suite you rely on next year.

Most security teams already have web application and API protection, or WAAP, on the roadmap. The category makes sense. Modern applications expose APIs everywhere, and WAAP was built to defend that surface with a bundled set of capabilities that typically includes WAF, bot management, DDoS protection, and API security. What gets less attention is where that protection stops, and what remains exposed on the other side of it.

If you ship compiled or interpreted code to client devices, that gap matters more than most WAAP conversations acknowledge. WAAP protects traffic moving between clients and servers. It does not protect the code once it lands on a user-controlled endpoint and can be inspected, modified, or repackaged. That is a separate attack surface, and it needs its own controls.

Key Takeaways:

• WAAP secures the network layer through WAF, bot mitigation, DDoS protection, and API security but does not protect code once it reaches client devices
• Adoption is growing quickly due to API expansion, cloud architectures, and compliance mandates like PCI DSS 4.0
• Client-side code remains highly exposed and is now easier to reverse engineer as AI tools lower the technical barrier
• Code-level protections such as obfuscation, anti-debugging, RASP, and tamper detection address this gap and extend security beyond the network boundary

What is web application and API protection (WAAP)?

Web application and API protection (WAAP) is a security category that combines four capabilities into a single platform: web application firewall (WAF), bot management, DDoS protection, and API security. It secures the network boundary between users and your applications. Code that runs on client devices sits outside WAAP’s inspection scope.

Gartner formalized the WAAP category around four required capabilities, each addressing a distinct attack class that a traditional web application firewall couldn’t handle on its own:

• Web application firewall (WAF): HTTP-layer policy enforcement against injection attacks, cross-site scripting, path traversal, and the OWASP Top 10. Modern WAFs layer machine learning over signature-based rules to catch novel patterns before a new rule ships.
• Bot management: Behavioral analysis, device fingerprinting, and challenge mechanisms that distinguish legitimate crawlers from credential-stuffing operations and scraping bots. A WAF doesn’t have the session-level visibility to do this reliably.

• Distributed Denial of Service (DDoS) protection: Volumetric attack absorption at the edge, inline, before traffic saturates your origin.

• API security: Discovery, schema enforcement, authentication validation, and detection of abuse patterns like broken object-level authorization and excessive data exposure.

The API security capability is what pushed the category beyond WAF entirely. WAFs, even next-gen ones, do a poor job protecting APIs because APIs don’t follow the request patterns WAFs were built to model. That’s not a configuration problem. WAF vendors have been trying to bridge that gap for years. WAAP emerged because WAF alone was not enough to address APIs, bots, DDoS, and modern cloud traffic patterns in a single runtime protection model.

NIST SP 800-228, published in June 2025 and updated in March 2026, references web application firewalls as part of API protection guidance for cloud-native systems, positioning them as components within broader API security architectures. That alignment can help security teams explain WAAP as part of a broader, standards-informed API protection architecture in procurement and audit conversations.

Why WAAP adoption is accelerating

WAAP adoption is accelerating for structural reasons. Microservices architectures mean a single application now exposes dozens of API endpoints, each one an attack surface a perimeter firewall wasn’t designed to model. Cloud migration dissolved the network boundary those firewalls relied on.

And regulation caught up: PCI DSS 4.0’s Requirement 6.4.2 elevates automated protection for public-facing web applications from a best-practice option to a defined requirement, mandating the use of a solution that continuously detects and prevents web-based attacks. More broadly, GDPR’s risk-based requirement for appropriate technical and organizational measures supports layered security controls, even though it does not prescribe WAAP specifically.

WAAP fits how security teams actually work now. Platforms that expose management APIs can be wired into CI/CD pipelines, letting teams enforce security posture at deploy time rather than bolting it on afterward. That’s what makes WAAP a workable component of a DevSecOps workflow rather than another tool that security owns and engineering routes around. That makes WAAP easier to integrate into modern cloud security workflows instead of treating it as a standalone perimeter appliance.

What WAAP doesn’t protect: Client-side code

WAAP inspects traffic between clients and servers. That is its scope. Once a compiled binary, packaged mobile app, or JavaScript bundle is delivered to a client device, it falls outside WAAP’s direct field of view.

That matters because client-side code can reveal much more than teams expect. .NET assemblies and Java bytecode can often be decompiled into readable pseudo-source. Android APKs can be unpacked, modified, and repackaged. JavaScript is delivered in a form that remains structurally understandable even when it is minified. From that access, attackers can extract business logic, map API behavior, identify enforcement checks, recover sensitive strings, and modify code paths for fraud or abuse. Network-layer protection does not stop any of that once the code is already in the attacker’s hands.

How AI tools have lowered the barrier to reverse engineering

The code-level threat is not new. What has changed is who can carry it out and how quickly. Reverse engineering used to require deep experience with assembly, decompilers, and compiler patterns. Recent LLM-based decompilation work has lowered that barrier by helping analysts reconstruct higher-level representations from binaries and by accelerating tasks like navigation, naming, and logic recovery.

That does not mean AI can perfectly reconstruct any protected application. It does mean unprotected or lightly protected code is easier to analyze than it was a few years ago. Research in this area consistently treats readable structure and preserved patterns as helpful context for LLM-based decompilation, which is exactly why obfuscation and harder-to-interpret binaries still matter. If your security posture stops at the network layer, your client-side code is exposed to a class of tooling your WAAP deployment will never see.

How code protection closes the WAAP gap

Code protection secures applications at the binary level, where WAAP cannot reach. The core controls are obfuscation, control flow protection, anti-debug, runtime application self-protection (RASP), and tamper detection. Together they address the two phases attackers work in: static analysis before the code runs and dynamic analysis while it runs.

Static analysis attacks the code before it runs. Code obfuscation and control flow protection address this phase directly. Obfuscation transforms compiled code at the instruction and structural level: renaming symbols, flattening control flow, inserting opaque predicates, encrypting strings. The binary executes identically. It just stops being readable. Control flow protection restructures execution paths to defeat static analysis tools that follow the program’s logic from entry point to output, which is most of them.

Dynamic analysis attacks the code while it runs. Anti-debug controls block or detect debugger attachment, the primary tool for live inspection. Runtime application self-protection (RASP) goes further, monitoring the execution environment for signs of tampering, injection, or unauthorized instrumentation. When it detects something, it responds before the attacker gets what they came for.

Tamper detection handles the redistribution threat separately. If an attacker repackages your APK, patches your .NET assembly, or modifies your JavaScript before execution, tamper detection identifies the modification at runtime. The response depends on your policy: degrade functionality, log the event, terminate the session, or some combination. The point is that you find out, and the attacker doesn’t get a clean run.

PreEmptive code protection tools: Dotfuscator, DashO, and JSDefender

PreEmptive has been delivering code-level application protection for over 25 years, across the platforms where client-side exposure is actually a problem.

Dotfuscator for .NET 

Dotfuscator for .NET provides code obfuscation, tamper detection, and runtime checks for .NET Framework and .NET Core applications. It integrates directly with Visual Studio and MSBuild, so protection applies at build time inside the pipelines your team already runs. It doesn’t require a separate security workflow, so protection doesn’t depend on someone remembering to run it.

DashO for Java and Android

DashO™ for Java and Android delivers code obfuscation, control flow protection, tamper detection, and runtime protections for Java applications and Android APKs. Android’s open distribution model means a repackaged APK with malicious modifications can appear in third-party stores within hours of your legitimate release. DashO makes that repackaging significantly harder and detectable at runtime.

JSDefender for JavaScript

JSDefender for JavaScript applies obfuscation and tamper detection to JavaScript running in browsers and Node.js. JavaScript is the one runtime where your code ships as near-source by definition. Minification helps with performance. It doesn’t protect your logic. JSDefender addresses the gap that most teams don’t think about until they find their algorithm running inside a competitor’s product.

WAAP Protects The Perimeter. PreEmptive Protects The Code.

WAAP is the right tool for securing the network boundary around your web applications and APIs. If you don’t have it deployed, that’s the first conversation to have. But for any application that ships compiled or interpreted code to client devices, the network perimeter isn’t the whole picture. The binary is a separate attack surface, and it needs its own defense.

WAAP and code protection operate on different layers. WAAP inspects traffic and blocks request-layer attacks against your servers. Code protection runs inside the application itself, defending against static analysis, reverse engineering, and runtime tampering after the binary ships. Teams that distribute client-side software need both.

PreEmptive provides that layer. With Dotfuscator, DashO, and JSDefender, you can apply code obfuscation, tamper detection, anti-debug, and runtime self-protection across your .NET, Java, Android, and JavaScript applications, integrated into the CI/CD workflows you’re already running. Start your free trial today.

---

FAQ