
CCE Technical Intelligence Briefing — Volume 20
Volume 20 of the CCE Technical Intelligence Briefing arrives during a period of significant activity across mobile device forensics, macOS artifact analysis, Windows threat-persistence research and the rapidly evolving use of artificial intelligence in investigative workflows. Cellebrite's Autumn 2026 Release is the dominant vendor story of the period, with the company's agentic AI platform Genesis at its center. Multiple articles — from Cellebrite itself, from Forensic Focus, and from independent commentators — examine how Genesis is changing triage and analysis in correctional settings and in high-profile criminal investigations. The University of Idaho murder case provides one of the most instructive case studies in recent memory: digital forensics proved premeditation not through what Kohberger's devices contained, but through what they did not, and through a single manual power-off event at 2:54 a.m. that the evidence showed was deliberate. The lesson — that an absence of expected data is itself evidence requiring explanation — is one the briefing returns to in several contexts this volume.
The macOS section is unusually rich, driven largely by the Eclectic Light Company's sustained coverage of macOS 27 Golden Gate and its forensic implications. Articles this period address Unified Log retention variability and how to measure it empirically, a SwiftUI focus bug affecting log-browsing tools, Spotlight indexing changes in Golden Gate, iCloud Drive sync timestamp generation, XProtect 5368 with nine new YARA rules covering four new malware families, and the Local Network privacy architecture change in 27.2 that finally allows apps to be removed from the access list. Two articles tackle the challenges the autumn DST transition creates for log-based timeline reconstruction, with LogUI 1.3 introduced to address one specific failure mode. Howard Oakley's retrospective on the Unified Log's history includes the High Sierra password-logging incident — a reminder that the same log that serves as a forensic record can itself capture sensitive secrets the user never intended to disclose. The Thesis Friday series contributes its second installment on Unified Log search term sets, introducing twelve multi-process sets each anchored to a specific investigative question and paired with explicit statements of what each set cannot prove.
The briefing's most technically ambitious material concerns WhatsApp Web browser forensics. Two closely related contributions from researcher Mihir Singh Choudhary — a detailed reverse-engineering analysis and an open-source extraction tool — establish the complete forensic picture of a WhatsApp Web browser profile. Contacts, metadata, thumbnails and media keys are all plaintext; the Noise device keypair is on disk wrapped with a fully disk-derivable key; message text alone is encrypted, using AES-CBC keyed from a salt that the WhatsApp server delivers transiently and never persists to disk. The practical result: metadata analysis is fully offline, reconnection using the on-disk credential enables message decryption, and WhatsApp Web's screen lock is the single setting that meaningfully changes the forensic picture. Elsewhere, the ORISA tool from Georgia Tech adds on-device AI conversation recovery to the Android mobile forensics toolkit; the ETag case study demonstrates a technique for proving file identity from a Chromium download record when the original file is gone; and leapps.org contributes a KeepSafe Android vault module for ALEAPP alongside updated Dropbox DBX decryption guidance and a thorough investigation into MAC time preservation in ZIP-based full file system extractions. The Windows threat-intelligence section covers WMI permanent event subscription persistence, ripgrep as a living-off-the-land binary, OneDrive artifact evolution, and an AI-compiled reference list of common malware archive passwords.
Prepared for Certified Computer Examiners (CCEs) and CCE candidates.
andreafortuna.org
Cellebrite's Advanced Unlocks and the case of the missing iPhone 17
A leaked Cellebrite Advanced Unlock datasheet lists a coverage range of roughly eight metres — a detail that makes no sense for a cable-based service and that prompted independent researcher Justin.sh to investigate the underlying technology. The post reviews the researcher's argument: Cellebrite's Advanced Unlock appears to involve a device that communicates wirelessly with the target iPhone over a short range, likely using an undisclosed RF interface that exploits a hardware or baseband vulnerability. The author separates what the analysis demonstrates — wireless proximity is required, the attack surface is not USB — from what it merely suggests, while noting that any unlocking service operating outside the USB port changes the forensic and legal calculus around chain of custody and reproducibility.
cellebrite.com
Genesis Momentum Powers Cellebrite's Autumn Release: More Evidence, Deeper Insight, Broader Access for Investigations
Cellebrite's Autumn 2026 Release centers on Genesis, its agentic AI solution, which investigators worldwide are calling a game-changer. The release expands device and operating system access and deepens analysis across evidence types. Investigators using Genesis report dramatic gains in what they uncover versus manual review alone. The platform transforms digital data from devices, cloud sources and third-party apps into defensible evidence while helping examiners validate findings and reduce review time. The release also broadens access to previously inaccessible data categories and adds new analytical workflows aimed at multi-source investigations.
cellebrite.com
Contraband Phone Triage: How Cellebrite Genesis Turns Device Backlogs into Investigative Intelligence
Contraband phone triage with Cellebrite Genesis allows correctional agencies to work across an entire batch of recovered devices simultaneously rather than reviewing them one at a time. Investigators ask questions across the batch, compare device identifiers and connections, and route the highest-risk devices to a human examiner first. The article explains why contraband phones matter beyond simple possession — they serve as command centers for extortion, drug trafficking and gang coordination — and covers how drones have become the dominant delivery mechanism. Genesis does not replace forensic extraction, chain of custody or examiner judgment; it accelerates the prioritization step so limited lab capacity is directed where intelligence value is highest.
cellebrite.com
Every Device Tells a Story: Turning contraband phones and drones into actionable intelligence
Cellebrite webinar (60 minutes including Q&A) for correctional and law enforcement teams dealing with contraband phone seizures. The session covers how phones, drones and outside accomplices form a single criminal network and why treating each seizure as an isolated incident leaves the network intact. Topics include what to ask of every recovered drone and contraband phone, how cross-device analysis surfaces accomplices and patterns, how AI-assisted triage helps small teams cut time from seizure to intelligence, and how to build a sustainable program when seizure volumes outpace staffing.
cellebrite.com
Lessons From the Forensic Examiners Behind the University of Idaho Murders Investigation
At 2:54 a.m. on the night of the University of Idaho murders, Bryan Kohberger's phone was manually powered off with Wi-Fi and cellular data disabled and the battery at 100%. On devices that otherwise yielded little, that single detail became what one examiner called the most damning evidence that the murders were planned. DNA identified the suspect; digital forensics established premeditation — largely through what was absent. The article details how an FBI examiner brought in two Cellebrite specialists when expected data was missing, the core lesson being that an absence of data is itself evidence that demands explanation, and the investigative steps taken to reconstruct Kohberger's digital footprint around the time of the killings.
cellebrite.com
How Digital Intelligence Is Transforming Contraband Phone Investigations in Correctional Facilities
Contraband phones are no longer fringe anomalies in correctional settings — they are embedded tools enabling extortion, drug trafficking, gang coordination and violence from inside facilities. Drones have become the primary delivery mechanism, and UAV data extraction has grown significantly more complex. Traditional forensic tools were designed for the laboratory, not for the speed and volume corrections environments require. Digital intelligence platforms now allow agencies to move from device intake to actionable insight in minutes rather than days, identifying threats before they materialise. The article covers extraction workflows, cross-device linking and how unified platforms support multi-agency collaboration on prison intelligence cases.
eclecticlight.co
Can I browse log entries from 2 weeks ago?
The macOS Unified Log does not delete entries by age; it deletes to control total size. Entries are written to tracev3 files across four subdirectories of /private/var/db/diagnostics, each with its own retention policy, making retention unpredictable and hard to measure precisely. The log command tool claims coverage of up to two weeks, but the author's empirical testing shows the actual window depends heavily on system activity and available storage. The article provides a methodology for measuring current log retention on any Mac, explains what the tool's self-reported timestamps mean, and sets realistic expectations for examiners who rely on Unified Log coverage as an investigative timeline anchor.
eclecticlight.co
Hero or hooligan: Death of Orpheus
An art-historical survey of the death of Orpheus in European painting. The post traces how painters across four centuries — Rubens, Gustave Moreau, Odilon Redon and others — depicted the same mythological moment, the maenads tearing Orpheus apart after his return from the underworld, and considers what each artistic choice reveals about the era's attitudes to heroism, transgression and the limits of art. Part of a recurring series on the eclecticlight.co blog examining classical myth through the history of visual representation.
eclecticlight.co
How Spotlight indexes in Golden Gate
Spotlight indexing was substantially revised starting in macOS Tahoe 26.6, which Apple described as preparing the index for macOS 27. Golden Gate 27.0 also fixed longstanding bugs in the indexing of plain text files. Using the author's free SpotTest diagnostic utility — which creates a standard set of test files and queries Spotlight against them — the article documents how indexing now behaves in Golden Gate 27.0.1 across plain text, RTF, PDF, HTML, DOCX and other formats. The findings are relevant to forensic examiners using Spotlight metadata as a source of file-access or indexing timestamps, since index behavior determines what is recorded and when.
eclecticlight.co
LogUI 1.2 fixes selection and focus in macOS 27
LogUI 1.1 had an embarrassing bug in macOS 27 Golden Gate: clicking a log entry selected it without moving focus, requiring Tab and Shift-Tab workarounds to fully highlight an entry. LogUI 1.2 build 91 resolves the issue using a workaround contributed by Philip Savoulidis, and is strongly recommended for anyone reading the Unified Log on Golden Gate. The update remains compatible with earlier macOS versions. For forensic examiners using the Unified Log as a primary evidence source, reliable selection and focus are essential for accurately capturing and referencing specific entries.
eclecticlight.co
Testing iCloud Drive with Cirrus 1.17
Cirrus 1.17 adds a timed iCloud Drive upload test: at the top of a minute, the tool creates a test file, triggers an upload, and measures exactly how long sync takes to complete. The test produces a Unified Log extract that can help diagnose sync failures or latency issues. The article walks through the test procedure and explains how to interpret the log entries generated during a sync cycle. For forensic examiners, iCloud Drive sync timestamps are a source of provenance and timeline data; understanding when and how sync activity is logged is prerequisite to relying on those timestamps in an investigation.
eclecticlight.co
Apple has just released updates to XProtect and XProtect Remediator
XProtect version 5368 and XProtect Remediator version 170 were released to all Macs. The XProtect update adds nine new YARA rules covering MACOS.CHEEKYCOBRA, MACOS.FLUSTEREDFLAMINGO, MACOS.SHADYSHOELACE and MACOS.TIMELYTURTLE malware families, and removes the rule for MACOS.DUBROBBER.CHBI. New Osascript rules cover MACOS.OSASCRIPT.SHCUPA with amendments to three existing rules. XProtect Remediator adds a new scanner for Flusteredflamingo. As with all XProtect updates, Apple releases no public information on the specific threats addressed.
eclecticlight.co
Local network privacy is getting better
macOS Sequoia added a Local Network privacy control, but unlike TCC-managed permissions it was implemented as a Network Extension packet filter, and there was no way to remove an app from the list once it appeared — only toggle it on or off. macOS Golden Gate 27.2, released the following week, adds the ability to add and remove apps from the Local Network privacy list. The article explains the architecture difference between Local Network privacy and standard TCC controls, why the limitation existed, and what the fix enables. For examiners, Local Network access records are a source of lateral-movement evidence; understanding how the control works and where it is logged is essential context.
eclecticlight.co
Why Golden Gate may not change focus correctly
macOS 27 Golden Gate introduced a SwiftUI focus bug where clicking a selectable item in a window selects it visually but does not move keyboard focus to it, so arrow-key navigation continues from wherever focus was previously. The article documents the bug with Finder examples and explains the underlying cause: a SwiftUI framework regression in how focus events are dispatched on click. The issue affects any SwiftUI-based application, including several forensic tools. The author connects this to the LogUI fix in the previous post and provides a workaround until Apple issues a patch.
eclecticlight.co
Reviewing the Desktop metaphor
The macOS graphical interface has changed dramatically in the forty years since the original Mac, but its underlying metaphor — a physical writing desk, with files, folders and a trash bin — has not. The article traces the metaphor from its origins at Xerox PARC in the early 1970s through the Lisa, the original Mac, the transition to color in System 7, Aqua and beyond, examining what has stayed constant and what has quietly shifted. For forensic practitioners, the Desktop metaphor directly shapes the artifact landscape: where files are expected to be found, what constitutes a user action versus an automatic system action, and how the system's model of 'the user's space' has evolved across macOS versions.
eclecticlight.co
Last Week on My Mac: The Wit and Wisdom of the Log
A curated retrospective of notable Unified Log entries from across macOS history. The article's most forensically significant example involves macOS High Sierra 10.13: when APFS was introduced, Disk Utility engineers unknowingly logged the full diskutil command including the user's plaintext passphrase for every encrypted volume creation. The log entry read as a full shell invocation including the -passphrase flag and its value. The example illustrates both how the Unified Log can capture sensitive secrets that users do not know are being recorded, and how Apple's internal log sanitisation rules can fail when new subsystems are added.
eclecticlight.co
How free space has been changing, with an update to DeltaFree
DeltaFree 1.17 adds CSV export of per-container free-space datasets, enabling trend analysis and charting outside the app. Using data collected across two months on a Mac mini M4 Pro, the author analyzes backup storage growth rate (approximately 7 GB per week) and projects remaining backup storage life. The article shows how the APFS container model — where internal and backup storage share a container pool — means apparent free space fluctuates as snapshots are created and pruned. For forensic purposes, understanding APFS container space accounting is relevant when interpreting volume metadata and when assessing whether a subject system had sufficient space for specific user activities.
eclecticlight.co
Apple has just released updates to Golden Gate 27.2, 26.7.2 and 15.8.2
Apple released macOS Golden Gate 27.2 (skipping 27.1), Tahoe 26.7.2 and Sequoia 15.8.2, alongside matching Safari security updates. The Golden Gate update download is 21.8 GB. The release notes include Siri AI improvements with wider language support, Messages contextual suggestions, Shared Albums improvements, and several accessibility and performance fixes. The security content for all three macOS versions addresses CVEs across kernel, Safari WebKit, and several system frameworks. For examiners, the update changes build numbers and may alter artifact locations or formats, particularly in Spotlight, iCloud and log subsystems.
eclecticlight.co
Daylight saving time, hieroglyphics, and LogUI 1.3
The autumn clock change exposes a fundamental ambiguity in Unified Log timestamps: the duplicated hour produces entries that cannot be unambiguously ordered without explicit time zone information. Tools using the log command (including Ulbow) receive entries in a compressed form — 714,000 entries squeezed into two seconds — because the tool does not resolve the time zone transition. LogUI 1.3, using the OSLog API directly, handles the transition correctly but still misreports the previous time zone label. The article documents the exact failure modes of each approach and the implications for examiners constructing timelines that span a DST boundary: without explicit zone attribution, timestamps in the duplicated hour are ambiguous evidence.
eclecticlight.co
Solve macOS update problems
A practical troubleshooting guide for macOS Software Update failures, covering the full range from updates that do not appear in the Software Update pane to glitches after an update completes. The article covers restart, Safe Mode (the procedure differs between Intel and Apple silicon), resetting the SoftwareUpdate preference domain, and when to involve Apple Support. For forensic examiners, failed or incomplete system updates can leave a Mac in a mixed-version state that affects artifact interpretation, particularly when update-related log entries are used to anchor a timeline or when Security Response content is inconsistent with the reported OS version.
forensicfocus.com
Investigating WMI Permanent Event Subscription Persistence
Windows Management Instrumentation (WMI) permanent event subscriptions are a well-documented but frequently overlooked persistence mechanism. Unlike Run keys or scheduled tasks, WMI subscriptions survive reboots and system reimaging, are not visible in standard autoruns output, and are stored entirely within the WMI repository at C:\Windows\System32\wbem\Repository. The article covers the three-component architecture (EventFilter, EventConsumer and FilterToConsumerBinding), how to query the repository for active subscriptions, what forensic artifacts survive after removal, and how to correlate subscription activity with Windows Event Log entries and Sysmon telemetry to establish when a subscription was created and whether it executed.
forensicfocus.com
Passware Kit 2026 v4 Simplifies BitLocker Decryption And Adds E01 Export
Passware Kit 2026 v4 extends BitLocker decryption support to additional AMD Zen/Zen+ systems — including Lenovo devices — and introduces the YellowKey method, which simplifies PIN recovery before decryption. The release also adds decryption of eCryptfs-encrypted Linux data and the ability to save decrypted full-disk encryption images and containers in E01 format, enabling direct ingestion into existing forensic workflows. Apple-related additions include password recovery for iWork 2026 files, a native Apple silicon binary in beta, and compatibility with macOS 27 Golden Gate.
forensicfocus.com
How Semantics 21 Is Changing The Way Investigators Work With Digital Evidence
Semantics 21 positions itself as a next-generation digital investigation platform that moves beyond evidence processing to evidence prioritization. The platform addresses the growing gap between the volume of digital evidence investigators face and the capacity of existing tools to surface what matters: investigators currently open files manually, watch footage in real time, and listen through audio recordings. Semantics 21 applies AI-driven analysis across victim identification, media review, CCTV footage, transcription and cross-evidence linking to route investigators to relevant material earlier in the process, reducing the manual review burden on complex cases.
forensicfocus.com
Magnet Forensics Launches Magnet AI Research Lab To Provide Peer-Backed Evidence To The Community
Magnet Forensics has launched the Magnet AI Research Lab, a dedicated program for peer-reviewed research on AI in digital forensics. The Research Lab's first published study, appearing in the Journal of Forensic Sciences, compared trained human examiners with an AI expert system on media authentication tasks. The AI scored 22 percentage points higher than the human average. The initiative is designed to give investigators, agencies and courts evidence — not vendor claims — to evaluate the accuracy and reliability of AI-driven tool outputs, addressing the growing question of how AI findings can be defended in court.
forensicfocus.com
End-To-End Digital Investigations: From First Access To Final Verdict
Forensic Focus webinar covering the full investigative lifecycle from initial device access through evidence analysis to court presentation. The session addresses the practical challenges examiners face at each stage: establishing a legally defensible extraction, managing evidence integrity across a multi-examiner workflow, presenting technical findings to non-technical audiences and withstanding cross-examination. The webinar is aimed at both junior examiners building a systematic approach and experienced practitioners looking to tighten their end-to-end process for high-stakes cases.
hexacorn.com
Solving practical problems with AI – Part 2 – the most popular, and most commonly used passwords for malicious archives
Malware samples distributed as password-protected archives almost always use a small set of well-known passwords. The author uses AI to compile and extend the standard list beyond the usual suspects (infected, malware, virus, inf3cted) to include values such as 1337, pass, god, virussign, 8466 and common numeric sequences. The post provides a machine-readable reference list of archive passwords used by threat actors and security researchers, useful for automated triage of archived malware samples. It also illustrates AI's value as an aggregation tool for tacit practitioner knowledge that is scattered across forums, tweets and blog comments rather than formally documented.
hexacorn.com
ripgrep as a lolbin
ripgrep's -z / --search-zip flag causes it to decompress any archive it encounters before searching, using external binaries listed in the tool's configuration: gzip, bzip2, xz, lz4, lzma, brotli, zstd and uncompress. Because the decompressor binary name is derived from the archive extension, an attacker can create a dummy archive (e.g. foo.gz) and place a malicious payload at the decompressor path (e.g. gzip.exe in the current directory or PATH). Running rg -z on that archive will execute the payload. This elevates ripgrep — a widely deployed developer and security tool — to living-off-the-land binary status, enabling execution without spawning a process that would look unusual in EDR telemetry.
leapps.org
Dropbox DBX Decryption, Nine Years Later
Nine years after the original research documenting the Dropbox DBX local database encryption scheme, the author revisits whether the method still works. The DBX file is Dropbox's local SQLite cache of file metadata, sync state and account information. The original decryption relied on a key derived from hardware identifiers. The post documents what has changed in the intervening years — including Dropbox client updates that altered the key derivation path — and provides an updated decryption method compatible with current Dropbox versions. For digital forensics practitioners, the DBX file remains a valuable artifact for establishing which files a user synced, when, and whether they were deleted from the cloud.
leapps.org
DFIR Thoughts: Put AI in the code, not the conclusions
The author argues for a disciplined division of labor between AI and the human examiner in DFIR work: AI belongs inside the tool — parsing artifacts, normalizing data, flagging anomalies — but not in the conclusions section of a forensic report. When an AI model generates an interpretive conclusion directly, the examiner cannot trace the reasoning, cannot cross-examine the logic and cannot defend it in court. By contrast, AI embedded in parsing code is a transparent, testable component of the tool's methodology. The post draws the boundary clearly: automate the data work, not the analysis, and document every AI-assisted step as you would any other tool decision.
leapps.org
Full File System extractions in Zip: MAC times, part two
The second part of an investigation into MAC time preservation in full file system extractions delivered as ZIP archives. Part one established that ZIP has two timestamp fields (MS-DOS and Extended Information, the latter carrying NTFS or Unix timestamps), and that many forensic tools populate only the MS-DOS field, losing sub-second precision and time zone information. This part tests additional tools and extraction paths, documents which combinations correctly preserve all three MAC times at full precision, and explains what an examiner must check before relying on ZIP-embedded timestamps as evidence. The findings have direct implications for timeline reconstruction accuracy in mobile device investigations.
leapps.org
KeepSafe for Android Caught an Intruder. It Was Me.
KeepSafe is a popular Android vault application that hides photos and videos behind a PIN. The author adds an ALEAPP module for KeepSafe, documents the app's SQLite database schema, and shows how the hidden media catalog can be extracted without the user's PIN from an acquired full file system image. The post covers the database location, the metadata structure (file hash, original path, vault entry date, deletion status), and what can and cannot be recovered if the vault was emptied. The addition to ALEAPP makes KeepSafe data accessible via the same automated parsing pipeline that handles dozens of other Android artifacts, lowering the barrier for examiners who encounter the app in casework.
magnetforensics.com
RDP: Forensic analysis of Windows Remote Desktop Protocol artifacts
Remote Desktop Protocol leaves evidence on both sides of a connection. On the source system, artifacts include recently used server entries in the registry (HKCU\Software\Microsoft\Terminal Server Client\Servers), Bitmap Cache files showing what was displayed on the remote desktop, and RDP connection logs in the Event Log. On the destination, Event ID 4624 (logon type 10) records interactive sessions; Event IDs 4778/4779 record session reconnections and disconnections. No single artifact tells the whole story: some record only an attempt, others confirm an interactive session. The article provides an artifact-by-artifact reference with locations, formats and what each does and does not prove.
magnetforensics.com
Introducing the Magnet AI Research Lab: Our first peer-reviewed study puts AI in digital forensics to the test
The Magnet AI Research Lab launches with a peer-reviewed study published in the Journal of Forensic Sciences comparing trained human examiners with an AI expert system on media authentication tasks. The AI scored 22 percentage points higher than the human examiners on average. The study provides the kind of independent, published validation that courts and accreditation bodies require before AI-driven outputs can be presented as forensic evidence. The Research Lab will conduct and publish ongoing peer-reviewed studies across AI applications in digital forensics, with findings publicly available to the community.
matthewplascencia.substack.com
Forensics is FIRE!: A Review of LAVA
LAVA (LEAPP Artifact Viewer Application) is Alexis Brignoni's GUI front-end for the full suite of LEAPP tools — iLEAPP, ALEAPP, RLEAPP and their kin — that brings all their outputs into a single unified interface. The author reviews LAVA from the perspective of a practitioner who has used individual LEAPP tools extensively, examining what changes when the outputs are combined: cross-artifact searching, unified timelines, and the ability to pivot between related artifact types without leaving the application. The review covers installation, the import workflow, and the practical differences between using LAVA as a daily driver versus running individual tools and reviewing outputs in a text editor or spreadsheet.
thesisfriday.com
From hits to a reconstruction: search term sets for the Unified Log (Part 2 of 2) – Thesis Friday #32
Part 2 of the Anchored Log Reconstruction series. Where Part 1 showed how individual Unified Log search terms can mislead, Part 2 provides the solution: search term sets that group hits from multiple processes within a defined time window to support a conclusion. A Face ID unlock on iOS 18.2.1 produces three log lines from three separate processes within 33 milliseconds, each recording a different stage of the unlock. The companion cheatsheet contains twelve such sets, each organised around a specific investigative question (screen unlock, app launch, network connection, and others), with a second column documenting what each set cannot show. Part of the author's ongoing series on building defensible timeline reconstructions from Apple log data.
Georgia Institute of Technology / Newswise
Deleted AI Chats May Leave Clues Behind
Georgia Tech researchers have developed ORISA, a tool that can recover deleted conversations from on-device AI assistants running on Android. ORISA can also reconstruct recovered fragments into a working AI session, restoring context that would otherwise be lost. The findings establish that deleting a conversation from a locally-running AI assistant does not necessarily remove every trace — remnants may persist in databases, cached files or unallocated space. The research adds a new artifact category to mobile forensics: on-device AI conversation history, which may be relevant in investigations where a subject used a local AI assistant to plan, research or communicate about criminal activity.
malwaremaloney.blogspot.com
OneDrive Evolution
An interactive visual reference documenting the evolution of the OneDrive local profile directory structure across client versions. The post presents a collapsible tree of the full OneDrive profile folder hierarchy, with SQLite databases highlighted in blue, JSON files in orange, and INI files in green. Version comparison mode shows added data in purple and removed data in red, enabling examiners to quickly identify what changed between OneDrive versions. For forensic practitioners, the reference maps the locations of OneDrive's SQLite sync databases, JSON metadata files and configuration stores — the primary artifacts used to establish which files were synced, when, and whether they were accessible to the user.
github.com / Mihir-Choudhary
Open-Source Tool Extracts WhatsApp Web Browser Data
WhatsApp Web Extractor (wawf) is an open-source command-line and GUI tool that reads WhatsApp Web's browser-stored data from an acquired Firefox or Chromium profile folder (Chrome, Edge, Brave, Opera), decrypts it, and produces an examiner-readable export. It supports Windows, Linux and macOS, and operates on Chrome/Edge/Brave (LevelDB/V8 serialisation) and Firefox (SQLite/structured-clone/Snappy) profiles without modification. The tool recovers contacts, chat and group metadata, message metadata, thumbnails and media keys from disk with no authentication, and optionally reconnects as the linked device — with explicit consent and legal-use warnings — to retrieve encrypted message text and on-demand history older than the browser's local cache.
medium.com / Mihir Singh Choudhary
Your Browser Is Keeping Years of Your WhatsApp. Here's What It Stores and How the Encryption Really Works
WhatsApp Web stores a large, mostly plaintext record on disk: contacts, chat and group names, message metadata spanning years, media thumbnails, per-message media decryption keys, and the device's Noise static keypair — all readable without any key. The only encrypted field is the message text, protected with AES-CBC keyed by HKDF-SHA256 of a root key plus a salt called companion_enc_static. That salt is delivered by WhatsApp's server inside the encrypted Noise login handshake and is never written to disk — meaning 100% offline decryption of message bodies is impossible by design. The root key and Noise keypair, however, are present on disk (the latter wrapped with a fully disk-derivable preliminary key), so a live-network reconnect using the on-disk credential will retrieve a fresh companion_enc_static and decrypt the full message history. Screen lock, if enabled, re-wraps the Noise key with a PBKDF2-stretched passcode and breaks the reconnect path from disk alone.
thelocalh0st.com / Sujay Adkesar
The Browser Kept a Receipt: Proving a Download Was the Same File with ETag
An investigation case study: a user downloaded a malicious ZIP from a URL that was still live at examination time, but the file was gone from the device — no Downloads folder copy, no quarantine record, no hash. The only survivor was the Chromium History SQLite database, which preserved the download URL and its server-assigned ETag. By re-fetching the URL and comparing the current ETag to the stored value, the examiner confirmed the file at the URL was the same file the user downloaded. The post explains the HTTP ETag mechanism, how Chromium stores it in the downloads table, and the conditions under which ETag comparison is reliable versus unreliable as an identity proof, including server-side caching layers that can serve stale ETags.