Atarus Learn
Engagement Reference
The notes we keep open during tests, one situation per page, arranged by what we test and by the stage of the kill chain. Each step has the command or request we run and the change that stops it. Use them only on systems you are authorized to test.
Tracks
Six marks per track, one for each stage: Recon, Access, Escalation, Persistence, Impact, Report. A filled mark has a page.
-
Network
3 pages
-
Web and API
4 pages
-
Cloud
1 page
-
Identity
2 pages
-
AI and LLM
2 pages
-
Social engineering
Not yet written
-
Physical
1 page
-
Reporting
1 page
Pages
14 pages
Network
-
External recon and subdomain enumeration
Map everything a domain exposes on the internet: names from DNS and certificate logs, dangling records, open ports and edge devices, each with its fix.
-
Internal pentest: Active Directory recon
Turn one domain account into a map of attack paths: password policy, unsigned SMB, LLMNR fallbacks, BloodHound paths and vulnerable AD CS templates.
-
Prove network impact without causing damage
Show what a compromised position can reach without harming it: counts instead of copies, screenshots of consoles, one marker file and a clean action log.
Web and API
-
Web app recon: the first hour of a pentest
Map a web app before attacking it: proxy every request, read the headers, pull API routes from JavaScript bundles and list every input worth testing.
-
Authentication and session testing
Find the weakest way into an account: username enumeration, session fixation, logout, JWT handling, password reset poisoning and code attempt limits.
-
IDOR and broken access control testing
Prove who can read and change what: swap object IDs, tamper with HTTP methods, call admin routes, and test mass assignment and tenant boundaries.
-
Prove a web finding's business impact
Prove what a web or API finding costs the business: one record, two test tenants, an agreed zero-value transaction and an impact paragraph that holds up.
Cloud
-
Cloud pentest: AWS and Azure IAM enumeration
Work out what a set of cloud credentials can reach: the identity, its escalation paths, secrets in functions and settings, public exposure and logging.
Identity
-
SPF, DKIM and DMARC: test email spoofing
Find out whether anyone can send mail as a domain: read DMARC, SPF and DKIM as a tester does, run one spoof test, and publish the records that close it.
-
Password spraying and MFA fatigue testing
Test password spraying and MFA push fatigue without locking anyone out, then put in the lockout, Conditional Access and KQL detections that stop both.
AI and LLM
-
LLM app recon: mapping an AI feature
Map an AI feature before testing it: where the model sits, which tools and data it reaches, its rate limits, its logging and the model behind it.
-
How to write up an AI or LLM finding
Write an AI or agent finding a product team can act on: exact input and output, severity by what the action could do, and controls that hold.
Physical
-
Badge cloning and physical access testing
Find out whether someone without a badge can get in: identify the card technology, test readers, doors and tailgating, and close each gap.
Reporting
-
How to write a pentest finding
Write a security finding an engineer can reproduce and fix without a call: impact title, severity reasoning, rerunnable steps, the fix, the retest.
No pages match. See the whole reference