In this section

Windows User Activity Forensics: Logon Types, ShellBags, LNK Files and Jump Lists

Module 0

Introduction

A Windows host records two different things about people. The Security log records accounts: which ones logged on, how, from where, and when they left.

The user's own registry hives and profile record what was done: which folders were opened in Explorer, which files were opened by which program, which networks the machine joined. Neither is enough alone. An account without activity says nothing about intent, and activity without a session cannot be tied to anyone.

Signs in08:07:28type 2Opens a share10:55:44ProprietaryOpens E:11:14:55removableOpens Dropbox11:30:22its own folderLogs off17:25:41session ends

jmorrison on 5 March 2026, from the Security log and ShellBags. The logon and logoff bracket the session; the folders say what was done inside it.

This lesson reads both for one day in this course's insider case: an engineer, two days from leaving the company, on his own workstation. It separates the logons that are people from the logons that are not, reconstructs one session, and reads what was opened inside it. Module 3 teaches each artifact in full, and ends with the hardest part of the topic, attribution.

01

Accounts on a Host

People, services and the logons in between

Every successful logon on a Windows host writes event 4624 to the Security log, whether it is a person typing a password, a service starting at boot, or another machine connecting to a share. Microsoft's documentation for the event lists the logon types that distinguish them, and they are the first thing to read in any account investigation. On the insider's workstation the counts by type are:

Logon type             All accounts   jmorrison
---------------------  ------------   ---------
2  Interactive                   36          36
7  Unlock                       152         152
3  Network                      119         119
5  Service                      212           0
10 RemoteInteractive              0           0

Service logons, type 5, are the largest group: 212 of them, every one for SYSTEM, NETWORK SERVICE or LOCAL SERVICE. They are Windows starting its own components and say nothing about a person. Removing them leaves the logons that might.

The account under investigation, jmorrison, has 36 interactive logons, 152 unlocks and 119 network logons over the period the log covers, and no remote desktop logons at all. That last absence matters. It means every session at a desktop on this machine was at its own keyboard, not over Remote Desktop from somewhere else, which narrows who could have been using it.

There are no type 11 cached logons either. Cached interactive means Windows checked the password against credentials it stored locally because no domain controller answered, so its absence says every one of these sign-ins was checked by the domain, with the machine connected to the corporate network at the moment the user signed in.

Failed logons belong in the same reading. There are 3 event 4625 failures on the host, all for jmorrison, all interactive, all a wrong password, and each on a different morning around 08:00: the pattern of someone mistyping a password at the start of the day rather than of an attack. Module 3 reads the failure reasons and the patterns that do indicate guessing.

02

Logon Types and What They Mean

Which ones put a person at the machine

The logon types sort into those that mean a person is present and those that usually do not:

1. Type 2, Interactivesomeone at the keyboard, typing a password or PIN2. Type 7, Unlockthe same person returning to a locked session3. Type 10, RemoteInteractiveRemote Desktop: a person, but not at this keyboard4. Type 3, Networka share, a printer, a management tool: no desktop5. Type 5, Servicea service started by the Service Control Manager

Which logons put a person in front of this machine. The top three are people; the bottom two are usually not, and they are most of what a Security log holds.

Type 2, interactive, is a person at the keyboard signing in. Type 7, unlock, is that person returning to a session they locked; Microsoft's table describes it simply as the workstation being unlocked.

Type 10, remote interactive, is a person signing in through Remote Desktop: a real session with a desktop, but not at this keyboard. Type 11, cached interactive, is a sign-in with credentials stored on the machine when no domain controller could be reached, which on a laptop usually means it was off the corporate network.

Type 3, network, is the most frequently misread. Microsoft describes it as a user or computer logging on from the network, which covers opening a share on this host, printing to it, a management tool querying it, or an antivirus console checking in. It produces no desktop and puts no person at this machine, even when the account named is a person's.

The other half of a session is its end. Event 4634 records a logoff and carries the logon type of the session that ended, and the Logon ID field that Microsoft notes can correlate an event with others from the same session ties the two together. A session is reconstructed by pairing a type 2 logon with its logoff and placing every unlock, process and file activity between them.

The fields beside the type say the same thing in a second way. Here are two logons for the same account on the morning of 5 March, a few minutes apart:

Two 4624 events, same account, same morning

Read the type, not the name

Interactive

08:07:28, type 2, from no address

Its process

User32, package Negotiate

Network

08:04:09, type 3, from 10.1.10.203

Its process

NtLmSsp, package NTLM

Same field

TargetUserName jmorrison in both

Reading

one is a person at this desk; the other is a connection

The interactive logon came through User32, the process behind the Windows sign-in screen, with no source address because nothing came over the network. The network logon came from another machine on the corporate subnet through NTLM.

The account name is identical, which is exactly why reading the name alone misleads: an investigator who counted every 4624 for jmorrison as "the user signed in" would put the user at the desk at 08:04:09, before they had arrived.

Reading types correctly is what stops two common mistakes: calling a network logon "the user signed in", and treating an unlock as a new sign-in rather than the same person coming back.

03

One Session, Reconstructed

The fifth of March

On 5 March, the day the case centers on, the Security log holds these events for jmorrison:

08:04:09  4624  type 3   10.1.10.203
08:07:28  4624  type 2   -
09:51:41  4624  type 7   -
11:21:33  4624  type 7   -
11:39:48  4624  type 3   10.1.10.23
12:00:04  4624  type 7   -
13:09:51  4624  type 7   -
17:04:36  4624  type 3   10.1.10.36
17:25:41  4634  type 2   (logoff)

The interactive session begins at 08:07:28 with a type 2 logon and ends at 17:25:41 with a logoff of a type 2 session. Between them, four type 7 unlocks show the person locking the machine and returning through the day, which is ordinary office behavior. The three type 3 logons, from other addresses, are network connections authenticating as the account, not visits to the desk.

The first row is worth a second look. The day's first event for the account is the network logon at 08:04:09, a few minutes before the interactive sign-in.

Some other machine on the network authenticated as jmorrison before the user sat down, which on a corporate network is usually a mapped drive reconnecting, a laptop dock or a management agent. It is not evidence of anything on its own, and it shows why the type 2 event, not the first event, marks the start of the session.

The export this course uses does not carry the Logon ID field, so the pairing here is made by type and order: the one type 2 logon and the one type 2 logoff. On a real host, matching the Logon ID removes the guesswork when several sessions overlap, and Module 3 does it that way.

That gives the frame for everything else on the day. Any folder opened, file accessed or program run between 08:07 and 17:25 under this account happened inside a session that began with someone typing jmorrison's credentials at this keyboard. Activity outside the frame, at night or after the logoff, would need a different explanation: a scheduled task, a remote tool, or another account.

The frame also bounds the question of who. No other account had an interactive or remote session on the host that day, so there is no second user whose activity could be confused with this one. Module 3 extends the reconstruction to every day the log covers and lines it up against the company's own records of when the person was in the building.

04

What Was Opened: ShellBags

Folders, in the order Explorer recorded them

Windows Explorer remembers how each folder a user opens should be displayed: its view, size and sort order. It keeps that memory in the user's registry, mostly in UsrClass.dat, in keys known as ShellBags. As a side effect, ShellBags record folders the user browsed, including on network shares and removable drives, and they persist after the folders themselves are gone.

The folders recorded on the morning of 5 March, in order:

Last written (UTC)   Folder
-------------------  ---------------------------------------------------------
2026-03-05T10:55:44  ngineering$\Projects\Active\NGE-Bridge-Phase2\Proprietary
2026-03-05T11:14:55  My Computer\E:\
2026-03-05T11:22:11  My Computer\E:\Engineering_Backup
2026-03-05T11:28:11  er\C:\Users\jmorrison\Dropbox\Export\export_2026-03-05.7z
2026-03-05T11:28:33  My Computer\C:\Users\jmorrison\Documents\Export
2026-03-05T11:28:44  My Computer\C:\Users\jmorrison\Dropbox\Export
2026-03-05T11:30:22  My Computer\C:\Users\jmorrison\Dropbox

Read in order, the morning tells a story without a single file being named. At 10:55:44 the user browsed into a folder named Proprietary on the engineering share. At 11:14:55 they opened the root of drive E:, then a folder on it called Engineering_Backup. Minutes later they were in an Export folder in Documents and an Export folder inside Dropbox, where Explorer recorded a 7-Zip archive named for the date.

ShellBags have limits. They record that a folder was viewed, not which files in it were copied. Their timestamps are when the registry entry was last written, which is close to when the folder was opened but not necessarily the first visit.

And they record folders the user opened, not folders a program created on its own. They point an investigator at where to look; the next two artifacts say which files.

The user's ShellBags also hold 89 distinct network share paths over the period, a map of the corporate data the account reached, which Module 3 compares with what the user's role needed.

05

Which Files: LNK Files and Jump Lists

Shortcuts Windows makes on its own

When a user opens a file through the shell, Windows creates or updates a shortcut to it in the user's Recent folder. These LNK files carry far more than a path.

The LECmd documentation lists the target's paths and timestamps, the volume and network share it was on, and the machine and MAC address that created the shortcut. On the insider's workstation, the Recent folder holds 83 shortcuts: 60 to the local disk, 15 to network shares, and 8 to a removable drive. The newest five removable ones:

Target accessed (UTC)  Drive      Serial     Target
---------------------  ---------  ---------  ------------------------------------
2026-02-20T13:15:08    Removable  B7E2-1A4F  tunnel_design_proprietary.dwg
2026-02-25T10:35:00    Removable  B7E2-1A4F  reinforcement_schedule_v4.xlsx
2026-02-25T10:39:32    Removable  B7E2-1A4F  soil_survey_results_MCR.pdf
2026-02-28T15:22:33    Removable  B7E2-1A4F  project_costings_internal.xlsx
2026-02-28T15:26:52    Removable  B7E2-1A4F  structural_analysis_final.xlsx

Every one points at E:\Engineering_Backup, and every one records the same volume serial number, B7E2-1A4F. Drive letters are reused by any USB device plugged in, but a volume serial identifies one formatted volume. The same removable volume held company files on four different days, beginning on 15 February. One morning's activity is now a pattern going back weeks.

Jump lists add the program. Each application pinned to or used from the taskbar keeps its own list of recent files, and the JLECmd documentation notes each jump list records the AppId of the program it belongs to, with full shortcut details for every entry. The 7-Zip File Manager's jump list on this workstation:

7-Zip File Manager   C:\Users\jmorrison\Dropbox\export_2026-02-15.7z
7-Zip File Manager   C:\Users\jmorrison\Dropbox\export_2026-02-20.7z
7-Zip File Manager   C:\Users\jmorrison\Dropbox\export_2026-02-25.7z
7-Zip File Manager   C:\Users\jmorrison\Dropbox\export_2026-02-28.7z
7-Zip File Manager   C:\Users\jmorrison\Dropbox\export_2026-03-02.7z
7-Zip File Manager   C:\Users\jmorrison\Dropbox\export_2026-03-05.7z

Six archives, each named for a date, each in the Dropbox folder, the last one on 5 March. Jump lists persist after the files they name are deleted, which makes them some of the most durable evidence of what a user worked with. Parsing the whole Recent folder is one command:

PS C:\> $p = "E:\ENG14\C\Users\jmorrison\AppData\Roaming\Microsoft\Windows\Recent"
PS C:\> .\LECmd.exe -d $p --csv E:\Out\ENG14 --csvf lnk.csv

The output is one CSV row per shortcut, with the target path, the target's own created, modified and accessed times, the drive type and the volume serial in separate columns, which is how the table above was built.

Sorting it by drive type and serial is the quickest way to find every file a user opened from a device that is no longer attached. The same command with JLECmd and the jump list folder gives the program-by-program view.

06

Where the Machine Was

Networks it joined

A laptop's location at the time of an action can matter as much as the action. Windows records every network it connects to in the SOFTWARE hive, under NetworkList profiles, with the date each profile was created and when it was last connected. The insider's workstation knows three:

Profile                 Category  Created               Last connected
----------------------  --------  --------------------  --------------------
NorthgateEng-Corporate  Public    2025-06-15T08:22:11Z  2026-03-05T08:01:22Z
Home-WiFi-5G            Private   2025-06-18T19:14:33Z  2026-03-04T22:33:44Z
Personal-Hotspot-5G     Private   2026-02-15T14:18:11Z  2026-03-05T11:20:33Z

The corporate network, a home Wi-Fi network created in June 2025, and a third profile named Personal-Hotspot-5G, created on 2026-02-15 and last connected at 11:20:33 on 5 March. A phone's hotspot is a way to take a corporate laptop off the corporate network without leaving the building, and the last connection to it fell in the middle of the morning the ShellBags describe.

The profile only says the network was joined, and when it was last joined.

How much data crossed it, and which programs sent it, is recorded elsewhere, in the System Resource Usage Monitor, which lesson 0.6 reads for exactly this host. Here it is enough to note that the machine's location changed during the session in a way the user chose, and that the change coincides with the removable drive and the Dropbox folder.

The category column adds a small detail. Windows marks the corporate profile Public and both of the others Private, which is the choice a user makes when Windows asks whether to let the machine be discoverable on a network. Someone chose to treat the phone's hotspot as a trusted network on the day it was first joined.

Module 3 reads the rest of the location evidence: the gateway addresses Windows records for each network, the time zone and clock history, and the cases where the network record is the only evidence that a device was somewhere at all.

07

Attribution

An account is not a person

Everything in this lesson ties actions to an account and a session. None of it, on its own, puts a particular person in the chair.

A shared password, an unlocked machine left unattended, a colleague who knew the PIN: each would produce exactly these artifacts with someone else at the keyboard. Attribution is the step from "this account, in this session" to "this person", and it needs evidence the machine cannot provide.

What the machine can provide is the strength of the account-level finding. On 5 March it is strong: one interactive session, opened with the account's credentials at this keyboard, no other session on the host, no remote access, and activity recorded in the account's own registry hives in a sequence that makes sense as one person's morning. Written down with its limits, it reads:

Attribution: 5 March, 10:55 to 11:30

Stated with its limits

Account

jmorrison: interactive session since 08:07

Session

no other interactive or remote logon on the host

Activity

Explorer folders in the user's own profile hive

Device

removable volume serial B7E2-1A4F, seen before

Gaps

no camera, no badge record: who was seated is inferred

Finding

activity in jmorrison's session, consistent with the user

Why the wording matters

Artifacts tie actions to an account and a session. Putting a named person at the keyboard needs evidence from outside the machine, and the finding says which it has.

The last row is deliberately careful. The record supports "consistent with the user" and stops short of "the user did it", because the artifacts that would close the gap are outside the machine: a badge record showing the person in the building, a camera, a conversation.

An investigator states which of those the finding has and which it does not, and leaves the conclusion about a person to whoever holds the rest of the evidence.

The same discipline applies in the other direction. If the badge record later shows the user was out of the building at 11:14, the artifacts do not change, but the finding does: activity in the account's session with nobody known to be at the desk is a different and more serious finding, about a shared or stolen credential.

Writing the account-level finding separately is what lets it survive new evidence about the person.

That wording is not timidity. Findings about people end up in disciplinary hearings, tribunals and courts, and a finding that claims more than its evidence supports damages the whole report when it is challenged.

08

What Module 3 Builds On It

Accounts, activity and attribution in depth

Module 3 takes each part of this lesson into depth: account forensics in the SAM and the profile list, authentication and session analysis across every logon type, ShellBags and LNK files, application usage, search history, geolocation and network history, Office document tracking, and attribution. Each lesson reads its artifact on real records and ends with a procedure for parsing it.

The habit to carry into it is the order this lesson used:

Module 3, in the order this lesson used

Module 3

Frame

Security log: logons, unlocks, logoffs, failures

Accounts

SAM and ProfileList: which accounts exist and when

Folders

ShellBags: where the user browsed, shares and drives

Files

LNK files and jump lists: which files, which program

Place

network profiles: which networks, when

Finding

about the account, with the person stated separately

Each row narrows the one before. The frame says when a person could have been at the keyboard; the account artifacts say which identities exist; the folder and file artifacts say what was done inside the frame; the network record says where the machine was while it happened.

A finding built in that order can be checked row by row, and a reviewer who disagrees with it can point at the row they dispute.

Start from the Security log and frame the session: who logged on, how, and until when. Then read the user's own artifacts inside the frame: folders, files, programs, networks. Then write the finding about the account, and say separately what is and is not known about the person.

Investigations that skip the frame tend to produce lists of artifacts with no story, and investigations that skip the last step tend to produce conclusions about people that do not survive a challenge.

The same frame is the backbone of the insider investigation module later in the course, where this host's full evidence set is read end to end, and of lesson 0.6, which follows the files from E: and the Dropbox folder to where they went.

Do

Practice

Who was at the keyboard, and what did they open?
prepared evidence

The prepared evidence below is the insider's 5 March: the account's logon events, the folders ShellBags recorded that morning, and the shortcuts to files on drive E:. Reconstruct the session, separate the logons that were not a person at the keyboard, and find what links the drive across days.

If you have a Windows machine of your own, run LECmd against your Recent folder and find a file you opened today from a USB drive or a share. Read the volume serial or the share path in its shortcut, then filter your own Security log for event 4624 and count your logons by type.

The Security log says which accounts logged on and how, and the logon type separates a person at the keyboard from services and network connections. Inside a reconstructed session, ShellBags, LNK files, jump lists and network profiles say what the account opened, where and on which device. All of it ties actions to an account; putting a named person in the chair needs evidence from outside the machine.

The next lesson follows an attacker rather than an insider: how they got in, and how they moved from one machine to the next.

Resources

Microsoft revises these pages; where a page and this lesson disagree, the page is newer.