Penetration testing for web, API and mobile.
Manual testing with a rate-limited toolkit, on assets you have signed off, in a window you choose. The output is not a scanner dump: it is a prioritised set of findings you can hand straight to a developer.
What can be in scope
Web applications
Single-page applications, server-rendered sites, customer portals and admin/back-office areas — including the parts that sit behind a login. Client-side storage, session handling, CSP, third-party widgets and the role boundaries between a normal user and an administrator.
APIs (REST)
Every endpoint the front end and the mobile apps talk to: authentication and token handling, object-level authorisation, mass assignment, rate limiting, error handling and information leakage, versioning and the endpoints that are live but undocumented.
Mobile apps (Android / iOS)
Static analysis of the shipped package — secrets left in the bundle, insecure local storage, exported components, deep links — plus dynamic testing of transport security, certificate pinning and root/jailbreak detection where a test device is available.
Configuration & infrastructure around it
TLS versions and ciphers, HSTS, security headers, CORS, CSP, DNS and mail authentication (SPF, DKIM, DMARC), CAA records, cookie flags, error pages, exposed build artefacts, cloud storage and database rules.
What gets tested
A catalogue, not a template: we agree at scoping which columns apply to your system. Each area below is a class of weakness, not a single check.
| Area | Typical questions that get answered |
|---|---|
| Authentication & session | Can accounts be enumerated? Is the OTP flow rate-limited and single-use? Password reset, e-mail verification, session fixation, logout invalidation, token lifetime and where tokens are stored client-side. |
| Authorisation (IDOR / BOLA) | Can one customer read or change another customer's data by changing an identifier? Are the admin endpoints enforced server-side, or only hidden in the front end? Does every object check ownership, not just authentication? Tested with two accounts, in pairs. |
| Injection & input handling | XSS (stored, reflected, DOM), SQL injection, NoSQL injection, SSRF, template injection, XML/deserialisation, file upload validation, path traversal, header injection. |
| Business logic | Price, quantity, discount or fee manipulation; coupon and referral abuse; state-machine bypass around a checkout; race conditions on balance, stock or booking; replay of a once-only action. |
| Payments & money flows | Where a payment intent is created and confirmed, which key ends up in the browser, whether amounts and currencies can be altered client-side, webhook signature verification, refund and subscription state changes. |
| Exposed secrets & artefacts | API keys and credentials shipped in JavaScript bundles, mobile packages or public repositories (including git history), source maps, configuration files, backup or admin paths that should not be reachable. |
| Client-side & dependencies | Token storage in localStorage versus secure cookies, missing CSP and clickjacking protection, postMessage and cross-window handling, out-of-support frameworks and known-vulnerable components. |
| Hardening & hygiene | HSTS and TLS configuration, security headers, cookie flags, CAA, DMARC/SPF/DKIM, response-code correctness, verbose error messages, stack traces and internal paths, logging and retention. |
| Third-party configuration | Keys with no referrer or API restrictions, permissive cloud storage or database rules, webhook endpoints accepting unsigned input, widgets loaded from third-party domains. Reported for you to fix or forward to the provider — never tested against the provider. |
What you receive
Report pack
- Owner-facing summary: what is at risk, what it would cost you, and in what order to fix it
- Developer-facing task list: one entry per finding with location, reproduction steps, evidence and a concrete fix
- Severity with a written justification and a confidence tag (confirmed / likely / possible)
- Evidence set: request and response pairs, screenshots, file hashes, timestamps
- Scope and limitations of the test, stated plainly — what was covered and what was not
- Retest result after your fixes, with the same evidence standard
Engagement options
- Fixed scope: one application or API on a signed asset list
- Project + retest: test, report, fix verification, closing statement
- Retainer: a test per release or per quarter, plus automation upkeep
- Pre-launch review: compressed timeline before a go-live date, with a call on critical findings
Where a staging environment exists, testing happens there. Where it does not, we agree production testing hours and a stop procedure before the first request.
How the testing is done
Reconnaissance starts from public information: DNS and certificate records, HTTP responses, served files, client-side bundles and app-store metadata. Then the manual work: mapping every request the application actually makes, testing the flows end to end with two test accounts, and following the paths that look under-protected.
Automated tooling is used where it is stronger than a human — crawling, parameter discovery, template-based checks, dependency and secret scanning — and its output is always verified by hand before it reaches your report. A tool that fires on a false positive costs your team more time than it saves.
Nothing is escalated beyond proof. If a finding means I can reach your data, the evidence shows the smallest possible demonstration on your own test data, not a copy of your database.
Toolkit
The same categories of tools the industry uses — from information gathering to reporting. Everything is rate-limited inside the window you approved, and no tool output lands in your report without being verified by hand.
Recon:
subfinder, amass, dnsx, httpx, nmap, crt.sh
Web:
Burp Suite, OWASP ZAP, ffuf, feroxbuster, nuclei, dalfox, sqlmap, interactsh
API:
Postman / Bruno collections, kiterunner, jwt_tool, mitmproxy, two-account authorisation testing
Mobile:
MobSF, jadx, apktool, Frida, objection
Code & dependencies:
semgrep, gitleaks, trufflehog, osv-scanner, npm audit, composer audit
TLS & mail:
testssl.sh, sslyze, sslscan, dig, DMARC report analysis
Method:
OWASP WSTG (web) and MASTG (mobile) as coverage checklists
$ subfinder -d example.com -silent | httpx -silent -td
$ nuclei -l live.txt -severity high -jsonl -o findings.json
$ ffuf -u https://app.example.com/FUZZ -w common.txt -mc 200,301,403
$ dalfox url https://app.example.com/?q=FUZZ --silence
— output verified by hand before reporting
Not included by default
These are deliberately excluded, and each needs a separate written addendum before it can happen.
- Denial-of-service, stress and load testing
- Social engineering, phishing and pretexting
- Physical access testing at your premises
- Red-team style persistence and covert channels
- Testing of third-party services (payments, cloud, chat, hosting, app stores)
- Reading, exporting or altering real customer data
If you need one of these
Tell me at scoping. Some of it I can arrange within a properly authorised engagement; some of it needs a different specialist or a written agreement directly with the third party involved, and I will say so rather than improvise.
A test that crosses a legal or contractual boundary is not a better test — it is a liability for you.