0. Orientation: what a forensic examiner is actually proving
A forensic tool does not create truth. It creates observations. Your job is to show how those observations were obtained, why they are reliable, what they support, and what they do not prove.
Five questions behind every finding
- Authority: were you permitted to collect and examine it?
- Identity: what exact device, file, account or stream is this?
- Integrity: has it changed since acquisition?
- Context: what does the artifact mean in this system?
- Reproducibility: could another examiner repeat the method?
Safety boundary for this manual
Labs use synthetic evidence, your own local files, simulated telemetry, or authorized sample images. Do not acquire third-party devices, bypass access controls, intercept traffic, or examine accounts without authorization.
Anti-forensics is taught as detection and interpretation, not as instructions for evading investigation.
0A. Global Quality, Standards & Forensic Reliability
The standard is not “the tool produced an answer.” The standard is: the method is appropriate, the tool behavior is understood, the evidence is traceable, the result is reproducible, and limitations are explicit.
Standards-aligned lifecycle
This manual uses NIST’s collection/examination/analysis/reporting model and the identification/collection/acquisition/preservation concepts reflected in ISO/IEC 27037. SWGDE guidance is used to reinforce contemporaneous documentation, evidence integrity, acquisition verification and examination discipline.
Scientific foundation
NIST’s scientific foundation review concludes that digital investigation techniques rest on established computer-science methods when appropriately used, while warning that evidence can be missed and artifact meaning can change as software evolves. That warning is built into every interpretation exercise here.
Publication quality gates
| Gate | What the learner should see | Failure condition |
|---|---|---|
| Authority & scope | Every practical activity states the evidence boundary and uses synthetic/owned data. | Instructions encourage collection from third-party systems or bypass of controls. |
| Integrity & provenance | Evidence IDs, hashes, transfer history, source context and acquisition logs are treated as separate concepts. | A hash is presented as proof of lawful origin, authorship or complete chain of custody. |
| Tool reliability | Known-answer tests, official syntax, logs, versions and limitations are recorded. | “The software said so” is accepted without test or corroboration. |
| Interpretation discipline | Observation → inference → alternative hypothesis → limitation. | Account/IP/timestamp evidence is converted directly into human intent or identity. |
| Reproducibility | Another competent learner can repeat the method and obtain the same deterministic lab result. | Unrecorded manual steps or mutable source evidence make repetition impossible. |
Lab 0 — Known-answer tool check
Verify your hashing environment before trusting it
Linux / GNU coreutilsKnown-answer testCreate a precisely defined file and compare your result to the published expected SHA-256 value. This tests the specific function you are about to use; it is not a blanket certification of the entire workstation.
mkdir -p "$HOME/forensics-lab/kat" && cd "$HOME/forensics-lab/kat" printf 'GNU Forensics KAT v1\n' > known-answer.txt sha256sum known-answer.txt # Expected SHA-256: # d930e915a541ace97c1b53399fc6715dbe3eac1186ec3593ea144cf16cee2794
Evidence notebook — practice contemporaneous documentation
Use fictional training identifiers only. This browser-side notebook never sends data anywhere; export it as JSON for your own lab record. Do not enter real case-sensitive information into a public training page.
Standards and primary documentation anchors
- NIST SP 800-86 — Guide to Integrating Forensic Techniques into Incident Response
- NIST — Digital Investigation Techniques: Scientific Foundation Review
- NIST CFTT — Computer Forensics Tool Testing and CFReDS reference datasets
- ISO/IEC 27037:2012 — identification, collection, acquisition and preservation of digital evidence (ISO lists this edition as current/under review at the time of this manual’s review)
- SWGDE current published documents, including digital evidence collection, computer forensic acquisition/examination, mobile and cloud guidance
- Wireshark/TShark official manual and official installation guide
- Volatility 3 stable documentation, official source repository and PyPI package
- India Code — Bharatiya Sakshya Adhiniyam, 2023
0B. The GNU Legal Learning Model: Learn → Practice → Defend
The public reference is intentionally useful on its own. Paid value must come from doing forensic work under constraints, not from hiding basic knowledge behind a paywall.
Public promise
You can learn the full conceptual method here without payment. No artificial content gaps are inserted merely to force a purchase.
Practitioner promise
The current Academy layer adds protected evidence sets, persistent case state, examiner challenge and exportable investigation records that cannot be reproduced by merely reading this page. Graded assessment and credentialing remain future gated features.
0C. GNU Forensics Starter Evidence Lab — Case S01
Move from reading about forensic method to examining a compact evidence set. Case S01 — Unexpected Outbound Transfer is a GNU-created, fully synthetic, authorized training case designed for roughly 45–75 minutes when completed with the investigation-method records. It teaches integrity verification, evidence inventory, working-copy discipline, contemporaneous examination logging, timeline reconstruction, corroboration, competing hypotheses, finding traceability, defensibility review and restrained reporting.
Published ZIP SHA-256: 24a26770790b072a6abd9803c26a22bc52705be1ad3b0d493a3c60c08e1508fc
| Record | Why it exists | Pack template |
|---|---|---|
| Evidence inventory | Makes every relied-upon source identifiable and hash-verifiable. | evidence_inventory.csv |
| Working-copy register | Shows which copy was analyzed and whether it matched the reference bytes. | working_copy_register.csv |
| Examination log | Captures tool/version, commands, results and problems such as AppArmor denial. | examination_log.csv |
| Timeline | Correlates events while retaining source and timestamp assumptions. | timeline.csv |
| Hypothesis matrix | Prevents the investigation from becoming a single-story confirmation exercise. | hypothesis_matrix.csv |
| Findings register | Forces each material claim to show observations, corroboration, alternatives and limitations. | findings_register.csv |
| Defensibility review | Tests whether another examiner can trace and challenge the conclusion. | defensibility_checklist.md |
| Mini-report | Converts technical work into a concise evidence-based professional finding. | mini_report.md |
What is inside
- Case brief and integrity manifest
- Synthetic Linux authentication and shell-history evidence
- Filesystem metadata
- Synthetic proxy telemetry
- Standard libpcap network capture for Wireshark/TShark
- Asset/time context
- Chain-of-custody, notebook and hypothesis templates
What you are trying to establish
- What account/session activity is observable?
- Was an archive created?
- Is there independent evidence of an outbound transfer attempt?
- Which sources corroborate one another?
- What remains unknown?
- What evidence would you seek next?
Starter workflow — verify → preserve → examine → correlate → report
Synthetic / authorizedLinux + optional TSharkRun these commands as your normal user from the directory containing the downloaded ZIP. Do not extract the pack with sudo. The reference copy stays untouched; derived notes and working files go under working/.
# 1) Verify the downloaded ZIP against the published digest. sha256sum GNU_Forensics_Starter_Evidence_Pack_v1.2.zip # Expected: # 24a26770790b072a6abd9803c26a22bc52705be1ad3b0d493a3c60c08e1508fc # 2) Extract and verify every supplied evidence/template file. unzip GNU_Forensics_Starter_Evidence_Pack_v1.2.zip cd GNU_Forensics_Starter_Evidence_Pack_v1.1 ./VERIFY.sh # 3) Preserve evidence/ as the reference copy; create a working copy. mkdir -p working cp -a evidence/. working/ # 4) Orient yourself before analysis. cat CASE_BRIEF.md cat INVESTIGATION_METHOD.md find evidence -type f -maxdepth 3 -print | sort # 5) Create editable case records without changing the supplied templates. mkdir -p working/records cp templates/* working/records/
Build a cross-source timeline
# Authentication / sudo activity grep -nE 'Accepted|session opened|sudo:|session closed' evidence/endpoint/auth.log # Shell-history commands (timestamp markers are stored as #epoch lines) cat evidence/endpoint/bash_history.txt # File metadata cat evidence/endpoint/file_metadata.csv # Proxy telemetry grep -n 'upload.example.test' evidence/network/proxy.log
Inspect the packet capture — preflight first
If TShark is installed, verify the tool and its ability to open the supplied PCAP before applying filters. The pack uses reserved documentation addresses and the .test namespace; it does not contact a real service. Offline tshark -r analysis should be run as your normal user; do not use sudo tshark for this lab.
# Tool + file-access preflight tshark -v test -r evidence/network/egress_capture.pcap && echo "PASS: shell can read PCAP" tshark -r evidence/network/egress_capture.pcap -c 1
# Beginner-safe fallback: create a temporary analysis copy. mkdir -p /tmp/gnu-forensics-s01 chmod 700 /tmp/gnu-forensics-s01 cp evidence/network/egress_capture.pcap \ /tmp/gnu-forensics-s01/egress_capture.pcap # Both hashes MUST match before analysis. sha256sum \ evidence/network/egress_capture.pcap \ /tmp/gnu-forensics-s01/egress_capture.pcap # Confirm TShark can open the analysis copy. tshark -r /tmp/gnu-forensics-s01/egress_capture.pcap -c 1
If the direct preflight succeeds, use evidence/network/egress_capture.pcap below. If you needed the fallback, substitute /tmp/gnu-forensics-s01/egress_capture.pcap. The temporary file is an analysis copy, not the reference evidence.
tshark -r evidence/network/egress_capture.pcap \ -Y 'dns.qry.name' \ -T fields -e frame.time -e ip.src -e dns.qry.name tshark -r evidence/network/egress_capture.pcap \ -Y 'http.request' \ -T fields -e frame.time -e ip.src -e ip.dst \ -e http.request.method -e http.host \ -e http.request.uri -e http.content_length
Why can TShark be denied when ls -l looks correct?
Linux may apply mandatory access controls in addition to ordinary user/group/mode permissions. For example, an AppArmor profile can deny a process read access even when the file owner can read the same path normally. If deeper diagnosis is needed, the learner pack includes TSHARK_PERMISSION_TROUBLESHOOTING.md. Do not weaken the reference evidence permissions to solve a tool-confinement problem.
Complete the investigation record before revealing guidance
Use the editable copies under working/records/. At minimum, identify each relied-upon evidence source, record material commands/methods, build the timeline, test at least two hypotheses and write one finding that cites its evidence IDs.
| Field | What good work looks like |
|---|---|
| Direct observations | Explicitly recorded facts tied to evidence IDs, not a narrative guess. |
| Inferences | Reasoned conclusions clearly separated from the source observations. |
| Alternative explanations | At least one plausible competing explanation tested against the evidence. |
| Attribution limits | Explains why account/session evidence does not by itself identify the human operator. |
| Transfer limits | Separates request, response, retention, content and downstream use. |
| Confidence | Calibrated and justified, with evidence that would change it. |
| Next evidence | Specific source(s) that would materially discriminate between hypotheses. |
Check your reasoning only after completing the timeline
Strong observations: the supplied logs record a successful analyst account SSH session, archive creation/hash/removal commands, a shell-history POST command, DNS resolution for the synthetic upload host, a TCP/HTTP request from 10.20.30.17 to 203.0.113.50:8080, and proxy telemetry for the same POST. These independent sources support a coherent event sequence.
Do not overclaim: this pack does not establish which human controlled the account, intent, authorization status, whether the filename accurately describes all transmitted content, or whether the remote service durably retained/used the body. A synthetic 201 Created response is another observation, not proof of downstream handling.
Good next-evidence requests: jump-host/identity/MFA records, EDR/process telemetry, filesystem/audit records, source-file hashes, remote service/object logs, authorization/change records, and relevant interviews.
1. Traditional Forensic Science Fundamentals
Crime-scene management
Establish a safe perimeter; separate people from evidence; record who entered; photograph or diagram device placement; note powered-on/off state, connected storage, visible screens, cables and network context before touching equipment.
Physical disciplines in context
Toxicology, forensic medicine, DNA/serology and trace evidence use different laboratory methods, but share principles familiar to cyber investigations: contamination control, validated methods, provenance, comparison against references and cautious interpretation.
Chain of custody ≠ a hash
A hash is an integrity check on bytes. Chain of custody is the documented history of possession, handling, transfer, storage and examination. A perfect SHA-256 value cannot tell you who held a drive overnight or whether the evidence was lawfully obtained.
Device ID, serial, location, state and collector recorded.
Evidence bag/seal ID recorded; power/network decision documented.
From collector to evidence custodian; both acknowledge transfer.
Working copy created; source preserved; hashes and tool/version logged.
Lab 1 — Chain-of-custody decision drill
No tools requiredInteractiveChoose the best first action for a powered-on laptop found with a cloud-sync client visibly transferring data.
2. Cyber Forensics Fundamentals
Computer architecture for examiners
CPU registers and caches are extremely volatile; RAM contains processes, decrypted content and transient credentials; persistent storage retains file-system and application artifacts; firmware and peripheral devices can retain configuration or logs. The OS mediates most of what the user and applications do.
Storage and file systems
| Concept | NTFS | FAT/exFAT | ext4 |
|---|---|---|---|
| Metadata anchor | MFT records, attributes, USN journal | Directory entries + allocation tables | Inodes, extents, journal |
| Deletion idea | Metadata may mark entry unused; clusters may persist until reused | Directory/allocation references change; content may remain | Directory/inode allocation changes; blocks may persist, but journaling/TRIM complicate recovery |
| Examiner caution | Alternate data streams, timestamp semantics | Weak metadata richness compared with NTFS/ext4 | inode reuse, journal behavior, mount-induced writes |
SSD caution: TRIM and controller garbage collection can make deleted-data recovery less predictable than on magnetic disks. Never promise recovery merely because a file was recently deleted.
Networking basics
A packet capture records packets visible at the capture point, not “the whole network.” NAT, TLS, VPNs, proxies, load balancers and asymmetric routing affect what an examiner can infer. Correlate PCAP with DNS, firewall, proxy, endpoint and application logs.
Three states of data
At rest
Files, databases, snapshots, backups, browser stores, disk slack, metadata.
In transit
Packets, flow logs, DNS, email transport, proxy records, API gateway telemetry.
In use
RAM, processes, sockets, mapped files, decrypted working data, tokens.
Across layers
A single user action can leave artifacts in all three states. Strong investigations correlate them rather than relying on one source.
3. The Digital Investigation Process
Bit-for-bit acquisition and verification
A forensic image attempts to reproduce addressable source data so analysis can occur on a copy. Hardware write blockers are commonly used with supported storage when appropriate; live, mobile, cloud and some modern storage workflows require different controls. Hashing is performed before/after or source/image as the workflow permits, with tool logs preserved.
Lab 2 — Safe acquisition + SHA-256 verification
Linux / GNU coreutils • synthetic imageThis deliberately images a file you create, not a real disk device. The commands below are validated for Linux/GNU coreutils. On macOS, use native equivalents such as shasum -a 256 and consult the local dd manual rather than copying GNU-specific options blindly.
mkdir -p "$HOME/forensics-lab/acquisition" && cd "$HOME/forensics-lab/acquisition" # Create a 4 MiB synthetic source image. It is disposable lab data. dd if=/dev/urandom of=evidence-source.img bs=1M count=4 status=none printf 'CASE=LAB-ACQ-001\nNOTE=synthetic evidence\n' | \ dd of=evidence-source.img bs=1 seek=65536 conv=notrunc status=none # Record source hash before acquisition. sha256sum evidence-source.img | tee source.sha256 # Create a bit-for-bit working image. dd if=evidence-source.img of=evidence-copy.img bs=1M conv=noerror,sync status=progress sync # Verify both byte streams have the same SHA-256 value. sha256sum evidence-source.img evidence-copy.img cmp --silent evidence-source.img evidence-copy.img \ && echo "PASS: byte-for-byte identical" \ || echo "FAIL: images differ"
Simulation — Evidence integrity failure
InteractiveClick “Acquire,” then deliberately alter the copy and verify again.
source: SHA256 2c7e…9a10 copy: [not acquired]
4. Computer / Disk Forensics
Assumes: file-system concepts from §2
Artifacts to correlate
File metadata, OS event logs, registry/config stores, browser databases, recent files, prefetch/application caches, shell history, recycle/trash metadata, persistence locations and user profiles.
Timeline thinking
Do not merely sort timestamps. Normalize time zones, record clock drift, distinguish filesystem/application/network time sources, and ask what process could have produced the sequence.
Lab 3 — Build and inspect a synthetic file timeline
Linux shellmkdir -p "$HOME/forensics-lab/timeline" && cd "$HOME/forensics-lab/timeline" rm -rf casefiles && mkdir casefiles printf 'quarterly notes\n' > casefiles/notes.txt sleep 1 printf 'invoice-id,amount\nA17,4200\n' > casefiles/invoice.csv sleep 1 cp casefiles/invoice.csv casefiles/invoice-copy.csv sleep 1 printf 'reviewed=yes\n' >> casefiles/notes.txt # Portable-ish metadata view (GNU find on Linux). find casefiles -maxdepth 1 -type f -printf '%TY-%Tm-%Td %TH:%TM:%TS %s bytes %p\n' | sort # Inspect one file in more detail. stat casefiles/notes.txt sha256sum casefiles/*
Learning objective: explain why modification time alone is insufficient to reconstruct a user story. On macOS, GNU find -printf is not present by default; use stat per file instead.
Timeline quality: normalize without destroying source context
Record the original timestamp, timezone/offset if known, system clock condition and the normalized comparison time (commonly UTC). Daylight-saving rules, application-specific timestamp formats, clock drift and provider-side time can change interpretation. Never silently rewrite a source timestamp.
Deleted files: what “deleted” really means
Most file systems separate a file's logical name/metadata from storage blocks. Deletion usually changes allocation/metadata references first. Recovery depends on whether metadata/content remains, whether blocks have been reused, whether encryption keys remain available, and whether SSD TRIM/garbage collection has removed content.
5. Network Forensics
Assumes: TCP/IP and packet-vs-flow distinction from §2
# Verify the tool first:
tshark -v
# TShark display-filter examples for an authorized capture file:
tshark -r capture.pcapng -Y 'dns' -T fields -e frame.time -e ip.src -e dns.qry.name
tshark -r capture.pcapng -Y 'http.request' -T fields -e ip.src -e http.host -e http.request.uri
tshark -r capture.pcapng -Y 'tcp.port in {80,443,8080}'Simulation — Streaming network triage
Real-time synthetic eventsStart the stream. Flag the sequence that most strongly suggests suspicious outbound activity. The simulator generates only fictional RFC 5737-style documentation addresses.
6. Memory (RAM) Forensics
Assumes: processes, parent/child, sockets
RAM can reveal active processes, injected code indicators, command lines, network objects and transient data not present on disk. Acquisition changes memory to some extent; document the acquisition tool, time and limitations.
Tool setup — Volatility 3
For learners, the simplest reproducible route is to install the official volatility3 Python package inside a virtual environment. The Volatility Foundation currently requires Python 3.8 or later and documents the installed command as vol.
# 1) Confirm Python is available. python3 --version # 2) Create an isolated environment for forensic tooling. python3 -m venv "$HOME/forensics-tools/volatility3-venv" source "$HOME/forensics-tools/volatility3-venv/bin/activate" # If the venv module is missing on Debian/Ubuntu/Pop!_OS, install python3-venv # with your OS package manager, then repeat the two commands above. # 3) Install the official Volatility 3 package from PyPI. python -m pip install --upgrade pip python -m pip install volatility3 # 4) Verify the installed CLI before using evidence. vol -h
Preflight the memory image before plugin triage
# Ask Volatility to identify/support the Windows memory sample first: vol -f MemDump.DMP windows.info # Then compare multiple process/network views: vol -f MemDump.DMP windows.pslist vol -f MemDump.DMP windows.pstree vol -f MemDump.DMP windows.psscan vol -f MemDump.DMP windows.netscan
Why windows.info first? The Volatility quick-start uses it to obtain OS/kernel information and check that the sample can be processed before moving into plugin-specific analysis. Individual plugins can have additional requirements; inspect them with vol <plugin> -h.
Where does vol.py come from?
vol.py is included in the official Volatility 3 source repository. It is not a separate file learners should search for or download from random mirrors. If you intentionally use a source checkout rather than the packaged CLI:
git clone https://github.com/volatilityfoundation/volatility3.git cd volatility3 # The project documents its stable branch as the latest stable code line. git switch stable # Create an isolated environment and install the project dependencies. python3 -m venv venv source venv/bin/activate python -m pip install --upgrade pip python -m pip install -e ".[full]" # From the repository root, vol.py is now present: python3 vol.py -h python3 vol.py -f /path/to/MemDump.DMP windows.info python3 vol.py -f /path/to/MemDump.DMP windows.pslist
A source checkout still needs the project’s Python dependencies. For ordinary learning, prefer the virtual-environment + PyPI installation above.
Official Volatility resources
Evidence/tool discipline
- Record the memory-image identifier and acquisition metadata.
- Record Volatility/Python versions and relevant plugin options.
- Preserve console output or export results according to your SOP.
- Corroborate anomalous processes/connections with other evidence.
Lab 4 — Synthetic process-tree triage
InteractiveNo memory image neededClick the process you would investigate first, then compare the explanation.
7. Mobile Forensics
Acquisition types
Logical: data exposed through OS/application interfaces. File-system: broader files/metadata where supported. Physical: lower-level acquisition where technically and lawfully possible. Availability varies widely by device and OS.
Common evidence
Call/message databases, app data, media metadata, backups, account tokens, location artifacts, notifications, browser data, Wi-Fi/Bluetooth history and SIM/eSIM-related metadata.
Lab 5 — Mock mobile backup artifact triage
Browser-only synthetic evidence| UTC | App | Peer | Artifact |
|---|---|---|---|
| 08:42:11Z | ChatApp | Project-Alpha | “upload complete” |
| 08:44:52Z | ChatApp | Project-Alpha | attachment: budget-draft.pdf |
| UTC | Source | Accuracy | Interpretation |
|---|---|---|---|
| 08:40:05Z | Wi-Fi geolocation cache | 120 m | Approximate, not GPS proof |
| 09:05:22Z | Photo EXIF (synthetic) | unknown | User-controlled metadata possible |
App version 7.2.1, last foreground 08:44Z, backup completed 09:12Z. Treat app metadata as one source to corroborate, not as independent proof of human intent.
8. Cloud, Database, Email & Web Forensics
Cloud
Audit/API logs, IAM changes, object-storage access, snapshots, flow logs, load-balancer logs, serverless logs, SaaS exports. Record account/tenant, region, retention settings and clock source.
Database
Transaction/audit logs, WAL/binlog/redo logs, query history, backups and application-layer records. Database forensic interpretation depends on engine/version/configuration.
Headers, transport hops, authentication results, server-side audit, mailbox actions, attachments and message IDs. Header IPs are contextual—proxies and hosted mail can change attribution.
Web/browser
History, downloads, cache, cookies, local storage, extension data, login/session artifacts and synced browsing data. Private browsing reduces some local artifacts but is not “forensically invisible.”
Lab 6 — Cloud audit log correlation
Browser-onlyEmail header reasoning: what can you safely conclude?
A Received: chain can show mail-transfer hops recorded by servers. SPF/DKIM/DMARC results help assess domain-level authentication. They do not automatically prove which human typed a message, and lower headers can be spoofed. Prefer trusted server-side copies and corroborating mailbox/audit evidence.
9. Malware Forensics & Anti-Forensics Detection
Static triage
Hash, file type, metadata, strings, imports, signatures, packer indicators. Static properties generate hypotheses; they do not prove runtime behavior.
Dynamic triage
Observe process, file, registry/configuration and network activity in an isolated lab. Record environment and snapshots so behavior can be reproduced.
Anti-forensics indicators
Unexpected timestamp discontinuities, secure-delete artifacts, missing logs where logs should exist, encrypted containers, steganography indicators, inconsistent metadata and tampered audit configuration.
Interpret cautiously
Encryption is normal security technology. A wiped file may be routine administration. An anomaly becomes meaningful through context and corroboration.
Simulation — Detect the anti-forensics clue
Timeline puzzle10. Advanced Techniques & Modern Cybercrime
| Domain | Useful evidence | Major caveat |
|---|---|---|
| Ransomware | EDR, process tree, ransom note, file changes, identity events, lateral movement, network egress, backups | Do not assume initial access from the encryption event alone |
| Cryptocurrency | Wallet addresses, transaction IDs, exchange records, device artifacts, communications | Blockchain addresses are pseudonymous identifiers; human attribution requires additional evidence |
| Dark-web activity | Authorized captures, seized-device artifacts, service metadata, communications | Jurisdiction, source reliability, operational security and lawful authority matter |
| Corporate espionage | DLP, file access, USB history, cloud sharing, source control, email, IAM | Access to a file does not by itself establish intent or exfiltration |
11. Legal Aspects, Ethics & Expert Testimony
Jurisdiction lens
Select a legal lens. Global is the default methodology view; the others are scoped orientation notes, not complete statements of local law.
Universal questions
- What lawful/organizational authority permits collection and examination?
- What is the exact scope and whose data may be affected?
- Can the exhibit be authenticated and its integrity/provenance explained?
- What disclosure, privacy, retention and minimization obligations apply?
- What expert-witness or report requirements apply in the relevant forum?
Method ≠ admissibility
A technically sound acquisition does not automatically make evidence admissible. Conversely, an admissibility rule does not validate the examiner's technical method. Document both the legal foundation and the forensic foundation.
India orientation
The Bharatiya Sakshya Adhiniyam, 2023 has been in force since 1 July 2024. Section 61 addresses electronic or digital records; Section 62 directs proof of electronic-record contents to Section 63; and Section 63 sets conditions for computer-output evidence and a certificate framework when such electronic record is submitted for admission. The statutory Schedule includes source/device particulars and hash information. The Information Technology Act, 2000 remains relevant to computer-related offences and other cyber-law questions.
Primary anchor: India Code, Bharatiya Sakshya Adhiniyam 2023, ss.61–63 and Schedule. Reviewed 24 Sep 2026.
England & Wales orientation
The Forensic Science Regulator's Code of Practice version 2 applies to forensic science activities in England and Wales and came into force on 2 October 2025. It establishes quality-standard requirements under the Forensic Science Regulator Act 2021 framework. Treat the Code as a quality/regulatory layer alongside the particular procedural and evidential rules governing the matter.
Primary anchor: UK Forensic Science Regulator / GOV.UK, Code of Practice v2.
United States — federal orientation
The Federal Rules of Evidence govern evidence in most federal proceedings. Rule 901 addresses authentication generally. Rule 902 includes self-authentication pathways for certain electronic records: 902(13) for certified records generated by an electronic process/system and 902(14) for certified data copied from an electronic device, storage medium or file using a process of digital identification. Authentication does not eliminate other objections such as relevance or hearsay, and criminal matters may raise additional constitutional/procedural issues.
Primary anchors: U.S. Courts current Federal Rules of Evidence; Rule 902 committee notes.
European Union — e-Evidence orientation
Regulation (EU) 2023/1543 creates European Production and Preservation Orders for electronic evidence in criminal proceedings and applies from 18 August 2026. Directive (EU) 2023/1544 supports the designated-establishment/legal-representative framework. This package focuses on obtaining and preserving cross-border electronic evidence; it does not replace each Member State's wider criminal procedure, evidence, data-protection and admissibility rules.
Primary anchors: EUR-Lex Regulation 2023/1543; Directive 2023/1544; European Commission e-Evidence guidance.
Expert-witness discipline
- Explain method before conclusion.
- Separate observation, inference and opinion.
- State tool versions, validation status and limitations.
- Avoid absolute claims when artifacts are probabilistic.
- Be able to reproduce exhibits and explain deviations.
Cross-examination stress test
- What could have altered this artifact?
- Why is this timestamp reliable?
- Who else had access?
- Was the relevant tool function tested?
- What evidence would contradict your hypothesis?
Expert testimony check
12. Customized Lab Generator
Choose the learner profile. The generator creates a compact practical path from the labs in this manual, including prerequisites and expected evidence outputs.
13. Capstone / Lab 8 — Real-time Incident Reconstruction
Scenario: an employee reports that a confidential PDF appeared outside the approved collaboration space. Your job is to reconstruct what the available synthetic evidence supports.
Investigator worksheet
13A. Free Case Zero — Evidence → Hypothesis → Defensible Finding
Scenario: At 09:17 UTC, monitoring detected outbound activity from workstation WS-17. Investigators are asked whether confidential material was exfiltrated. Your job is not to confirm the allegation; your job is to determine what the supplied evidence supports.
Evidence tray
Examiner reasoning
Examiner checkpoint — what would strengthen the case?
- Provider-side object upload/transfer records tied to the exact object and bytes transferred.
- Endpoint/browser/process evidence tying archive creation and transmission activity together.
- Reliable account/device/session attribution evidence.
- Independent evidence showing whether the externally shared object was accessed or downloaded.
Key lesson: a USB event, ZIP creation, network volume and share-link creation may collectively support a data-movement hypothesis, but each has different attribution and causation limits.
14. Reporting: from technical artifact to defensible finding
Recommended structure
- Authority, scope and questions
- Evidence received and identifiers
- Chain-of-custody references
- Tools, versions and methods
- Examination results
- Analysis/findings
- Limitations/uncertainties
- Exhibits + hashes
Language discipline
Observation: “The audit log records…”
Inference: “This sequence is consistent with…”
Limitation: “The log does not identify the physical person operating the account.”
Lab 7 — Turn an overclaim into a forensic finding
Writing labOverclaim: “The suspect definitely stole the file at 10:45.”
Show one defensible rewrite
“Endpoint file-access logs record account jdoe reading budget-draft.pdf at 10:44:51Z. A cloud audit record shows the same account creating an external share link at 10:45:13Z. These records support that the account performed the two actions in sequence. The available evidence does not, by itself, identify the physical person operating the account or prove that the file was subsequently downloaded by an external party.”
15. Competency Check — Can You Work Like an Examiner?
Use this rubric after the capstone. The goal is observable competence, not memorization.
Explain
You can distinguish evidence from interpretation, explain chain of custody versus hashing, describe data at rest/in transit/in use, and identify why live acquisition changes system state.
Perform
You can execute the synthetic acquisition/timeline labs, validate a known-answer hash, use documented TShark/Volatility syntax on authorized training evidence, and maintain an activity log.
Defend
You can connect each conclusion to an artifact and method, state an alternative hypothesis, identify missing evidence, explain tool limitations, and reproduce an exhibit.
Exit challenge — six questions
16. Lab Verification, Validation & Publication QA
| Lab | Validation approach | Expected result | Dependencies |
|---|---|---|---|
| 0 Known-answer hash | Creates a fixed 21-byte training file and compares SHA-256 with a precomputed reference digest | Exact digest d930…2794 | Linux/GNU coreutils |
| 1 Chain of custody | Logic/decision exercise | Document first; preserve volatility and authority | None |
| 2 Acquisition/hash | Runs only on a synthetic 4 MiB file; source/copy compared by SHA-256 and cmp | Matching digests + PASS | POSIX shell, dd, sha256sum, cmp |
| 3 Timeline | Creates/edits local lab files; prints metadata and hashes | Chronological metadata suitable for discussion | GNU find on Linux; stat/sha256sum |
| 4 Memory triage | Browser synthetic process table; real-image workflow now includes official-package installation, CLI verification, windows.info preflight and documented Volatility 3 plugins | Suspicious parent/child + network context identified; real-image path begins only after tool/sample preflight | None for simulator; Python 3.8+ and Volatility 3 for a real authorized image |
| 5 Mobile | Static synthetic backup artifacts | Correlated but cautious interpretation | Browser |
| 6 Cloud | Deterministic JSON-style event correlation | IAM → object access/share sequence | Browser |
| 7 Reporting | Overclaim reduction exercise | Observation/inference/limitation separated | None |
| 8 Capstone | Multi-source synthetic stream + report self-check | Timeline + alternative hypothesis + missing evidence | Browser |
| Starter Evidence Pack S01 | GNU-created synthetic multi-source evidence pack; ZIP + internal manifest hashes verified; PCAP structure/parser checked; v1.2 adds evidence inventory, working-copy register, examination log, timeline, hypothesis, findings, defensibility and mini-report records | Reproducible account/archive/network investigation with traceable findings, competing hypotheses and explicit attribution/transfer limitations | sha256sum/unzip; TShark optional for packet analysis |
| Case Zero | Deterministic browser reasoning exercise; selected evidence + four reasoning fields drive bounded readiness feedback | Evidence selection cannot substitute for attribution; alternative explanation and limitation required for strongest result | Browser |
Validation levels used by this edition
| Level | Meaning in this manual | Status |
|---|---|---|
| Content verification | Procedural/technical claims checked against primary or authoritative sources. | PASS |
| Syntax verification | Shell/CLI examples checked for documented syntax; JavaScript syntax parsed separately. | PASS |
| Known-answer execution | Synthetic acquisition/timeline labs executed; fixed hash known-answer included; Starter Evidence Pack v1.2 ZIP/manifest integrity and packet structure validated; investigation-method records added; TShark file-access preflight and verified-analysis-copy fallback retained. | PASS |
| UI structural validation | Internal anchors, duplicate IDs and key interactive controls checked. | PASS |
| Browser behavior | This release’s HTML structure and inline JavaScript syntax were validated at build time. Headless Chromium did not complete reliably in the build environment, so no browser-render PASS is claimed for these exact v1.3.13 bytes; the deployment URL and key interactive controls must be smoke-tested in a normal browser after publication. | PENDING — DEPLOYMENT SMOKE |
| Real forensic tool casework validation | Not claimed. Laboratories must validate their own tool/version/configuration and procedures for intended use. | ORG RESPONSIBILITY |
What this manual deliberately does not claim
- It is not laboratory accreditation, examiner certification, legal advice, a substitute for organizational SOPs, or validation of every version of any commercial/open-source forensic tool.
- Simulations teach reasoning; they do not reproduce every artifact, race condition, encryption state, provider behavior or device-specific acquisition constraint.
- Forensic conclusions remain dependent on the facts, collection method, tool behavior, source-system version and jurisdiction of the actual matter.
Reference anchors used to validate technical/legal statements
- NIST SP 800-86 — forensic techniques and the collection/examination/analysis/reporting lifecycle.
- NIST Digital Investigation Techniques scientific foundation review — reliability and limitations of digital investigation techniques.
- NIST CFTT/CFReDS — tool testing methodology and documented reference datasets for validation/proficiency use.
- ISO/IEC 27037:2012 — identification, collection, acquisition and preservation of digital evidence.
- SWGDE current guidance — digital evidence collection, acquisition, examination, mobile/cloud evidence and tool-testing expectations.
- Wireshark/TShark official documentation — installation, -r capture-file reading, capture vs display filters and -Y display-filter semantics.
- Volatility 3 stable documentation, official repository and PyPI metadata — Python requirement, installation, vol CLI, windows.info preflight, plugin model and source-checkout vol.py provenance; Volatility does not perform Windows memory acquisition.
- India Code — Bharatiya Sakshya Adhiniyam, 2023, Sections 61–63 and Schedule; reviewed 24 Sep 2026.
- SWGDE 10-Q-002-3.0 — training, competency, continuing education and continual assessment expectations.
- ISO/IEC 17025:2017 — competence, impartiality and consistent laboratory operation; current edition confirmed by ISO.
- SWGDE 19-Q-001 — QMS selection/implementation concepts for Digital & Multimedia Evidence organizations.
- SWGDE 18-Q-001-2.1 — baseline forensic tool testing and documentation.
- CASE/UCO — structured representation, provenance and investigative-action traceability.
- Forensic Science Regulator Code of Practice v2 — England & Wales quality requirements, in force 2 Oct 2025.
- U.S. Federal Rules of Evidence — current rules; electronic-record authentication orientation under Rules 901/902.
- EU Regulation 2023/1543 — European Production/Preservation Orders, applicable 18 Aug 2026.
17. Forensic Quality Management & Examiner Competence
Professional reliability comes from a managed system, not from an individual examiner being careful. ISO/IEC 17025 is an international competence standard for testing/calibration laboratories; SWGDE guidance adapts quality-system concepts to digital and multimedia evidence organizations whether or not they pursue accreditation.
The forensic quality loop
QMS maturity drill
18. Tool & Method Validation Laboratory
NIST CFTT and SWGDE both emphasize testing forensic-tool functions against defined requirements and known or independently checkable results. The goal is confidence in a specific function under specified conditions—not a blanket declaration that an entire tool can never fail.
Validation vs verification
Terminology varies by organization. Operationally, ask two questions: Is this method suitable for the intended purpose? and Does this installed tool/version/configuration perform the tested function as expected? Document your organization's chosen terminology.
When to reconsider testing
New tool/version, new OS/platform, material configuration change, new function, unexpected result, changed evidence format, changed workflow, or a risk assessment indicating prior testing may no longer be representative.
Validation record template
Lab 9 — Browser known-answer test
The fixed UTF-8 training string is:
GNU-LEGAL-DFIR-KAT-001|evidence-integrity|2026
Expected SHA-256
This validates only the browser's SHA-256 calculation against this known answer. It does not validate acquisition, provenance, or another forensic tool.
What a defensible test record contains
| Field | Why it matters |
|---|---|
| Requirement / intended use | Defines the exact capability that must work. |
| Environment | Tool version, OS, dependencies, hardware/configuration can affect behavior. |
| Ground truth | Lets the examiner compare observed output to a result known independently of the tool. |
| Observed output | Preserves what actually happened, including failures. |
| Limitations | Prevents the test from being interpreted more broadly than justified. |
| Disposition | Records whether the function is accepted, restricted, retested or rejected. |
19. Advanced Enterprise, Cloud, Container & AI Forensics
Modern investigations rarely stay on one disk. Enterprise evidence spans endpoints, identity providers, cloud control planes, SaaS, containers, orchestration, source control and AI-agent execution. The forensic challenge is preserving provenance while correlating different clocks, schemas and retention periods.
Windows evidence map
- Event Logs and relevant audit channels
- Registry hives and user/system configuration
- NTFS metadata and filesystem timelines
- Execution/persistence artifacts such as services, scheduled tasks and OS-version-dependent execution traces
- Browser/application databases and EDR telemetry
Linux evidence map
journald/syslog-family records and authentication logs- SSH, sudo, PAM, audit framework and service logs where enabled
- shell history—with caution about disablement, truncation and attribution
- package, persistence and service-manager state
- filesystem metadata, process/network telemetry and application logs
macOS evidence map
- APFS/filesystem metadata and snapshots where available
- Unified Logging and application-specific data stores
- LaunchAgents/LaunchDaemons and other persistence context
- browser, quarantine/download and user-activity artifacts
- device/security state and OS-version-specific privacy controls
Cloud / SaaS evidence map
- control-plane audit events
- identity and token activity
- object access/version history
- network/security telemetry
- snapshots/export metadata
- provider ticket/request identifiers and export method
Kubernetes / container evidence map
- Kubernetes API audit logs
- workload, pod, node and namespace identity
- container runtime logs and image digest/provenance
- deployment/ReplicaSet/DaemonSet history and configuration
- service-account/RBAC activity
- node-level evidence where lawful/available
Identity evidence map
- SSO sign-in events, MFA, conditional access and device context
- OAuth grants, tokens, service principals/workload identities
- role/group changes and privilege elevation
- session lifecycle, source network and risk events
AI / agent evidence map
- prompt/input and system-policy version
- model/provider/version and inference parameters where retained
- RAG retrievals, source document IDs and vector-store provenance
- agent plan/tool-call sequence and tool arguments/results
- API gateway/network logs and external effects
- human approvals, overrides and identity/session context
Enterprise correlation drill
14:02:11Z — MFA success for account svc-mlops from managed runner.
14:04:09Z — service account creates Job report-export.
14:05:22Z — agent tool call requests object legal-hold/summary.csv.
14:05:24Z — object read by workload identity.
14:05:31Z — runner establishes TLS session to approved vendor endpoint.
No retained tool-response body and no vendor-side receipt.
What is the strongest defensible statement?
20. Forensic Reasoning, Uncertainty & Cognitive Error
Digital evidence is rich enough to create convincing stories. Professional reasoning resists story-first analysis: observations are recorded before conclusions, alternative hypotheses are kept explicit, and the examiner asks what evidence would distinguish them.
The authenticated user deliberately moved confidential data.
An approved or misconfigured sync process moved the data without deliberate user action.
Another actor or malware used the user's authenticated session.
The apparent transfer is a logging, parsing, clock, duplication or artifact-semantics issue.
Competing-hypothesis exercise
Classify how each evidence item bears on the hypotheses. There may be more than one reasonable interpretation; the review focuses on whether you avoid treating weak/contextual evidence as dispositive.
| Evidence | Your classification | Reasoning cue |
|---|---|---|
| User MFA succeeds from a corporate device | Authentication supports account/session context; it does not establish the physical operator. | |
| Unsigned sync process launched by scheduled task created months earlier | Automated execution is directly relevant to an automation hypothesis. | |
| EDR records remote process injection into the user's browser before the transfer | Compromise evidence may weaken a simple user-intent attribution. | |
| Parser output differs from raw log timestamp by one hour around DST transition | Tool/time interpretation may affect event ordering. |
Bias controls
Observation → inference ladder
21. Global Professional Reference — Verification Matrix
| Quality dimension | What v1.3 does | Boundary |
|---|---|---|
| Primary-source alignment | Maps core practice to NIST, ISO/IEC, SWGDE, CASE/UCO concepts and primary legal sources for the jurisdiction lenses. | Standards/law can change; organizations must maintain controlled references. |
| Known-answer testing | Includes deterministic SHA-256 known-answer test and previously validated safe acquisition/timeline labs. | Does not validate every tool/version/function. |
| Reasoning quality | Requires alternatives, limitations, falsification and restrained attribution. | Automated self-checks are educational, not expert peer review. |
| Enterprise applicability | Extends evidence mapping to Windows/Linux/macOS/cloud/Kubernetes/identity/AI-agent systems. | Artifact semantics remain platform/version/provider specific. |
| Legal scope | Provides global method plus India, England & Wales, US federal and EU e-Evidence orientation. | Not local legal advice; laws outside these lenses require separate review. |
| QMS awareness | Introduces competence, controlled methods, tool testing, review, nonconformance and continuing improvement. | Not accreditation and not an organizational quality manual. |
| Tool onboarding | Explains command provenance and verification for GNU hashing utilities, TShark and Volatility 3 before tool-dependent examples are used. | Package availability, acquisition tooling and casework validation remain platform/organization specific. |
Reference anchors added in v1.3
- ISO/IEC 17025:2017 — laboratory competence, impartiality and consistent operation
- SWGDE — QMS for Digital & Multimedia Evidence organizations
- SWGDE — Minimum Requirements for Testing Tools
- NIST CFTT
- CASE — structured cyber-investigation information and provenance
- England & Wales — Forensic Science Regulator Code v2
- United States — current Federal Rules of Evidence
- European Union — Regulation (EU) 2023/1543 e-Evidence
22. Continue to the GNU Legal Digital & Cyber Forensics Practitioner Lab
Keep this global professional reference free on legal.gnugroup.org. Deliver protected practitioner cases, saved progress and Razorpay entitlement on learn.gnugroup.org, where the Academy infrastructure already belongs. Graded assessment and certificate issuance remain gated until the practitioner workflow completes end-to-end validation.
Global Reference
- Complete conceptual reference
- Standards and quality anchors
- Guided synthetic labs
- Interactive simulations
- Evidence notebook
- Case Zero reasoning drill
- Learning Coach + voice
- Self-check competency rubric
No account, payment or server-side learner state is required.
Practitioner Experience
- Three realistic investigation cases
- Guided → Analyst → Examiner modes
- Persistent Evidence Workbench
- Timeline reconstruction studio
- Hypothesis / alternative explanation review
- Senior Examiner challenge prompts
- Exportable investigation records
- Assessment and certificate follow after end-to-end validation
Current and planned practitioner case roadmap
| Status | Case | Primary evidence | Competency |
|---|---|---|---|
| CURRENT | 01 — Suspicious USB / data removal | Windows device-install/USB artifacts, filesystem, identity and endpoint evidence | Separate activity from attribution |
| CURRENT | 02 — Linux server compromise | journal/auth, SSH, web logs, filesystem, network | Reconstruct initial access and post-compromise activity |
| CURRENT | 03 — Suspicious data transfer | PCAP, DNS/TLS metadata, proxy and endpoint evidence | Correlate transfer indicators without overclaiming content |
| PLANNED | 04 — Memory investigation | process tree, command lines, connections, modules | Reason from volatile artifacts and tool limitations |
| PLANNED | 05 — Cloud identity compromise | identity, MFA, API/object audit and source context | Separate account activity from human attribution |
| PLANNED | 06 — AI agent incident | prompts, RAG retrieval, model/version, tool calls, API logs | Trace human → model → agent → tool → external effect |
| PLANNED | 07 — Unknown capstone | mixed evidence with distractors | Plan, investigate, document, defend and report independently |
/forensics/index.html + /forensics/GNU_Forensics_Starter_Evidence_Pack_v1.2.zip • static HTML/CSS/JS + downloadable synthetic ZIP • no external runtime dependency • no payment SDK • no API keys • browser-local notebook only • outbound links only to authoritative references and GNU Learning Academy.
Suitable for the existing GNU Legal static Docker/Caddy deployment model.