Anti-Forensics Field Manual

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== Run Keys ======

===== Definition =====
Run Keys are a Windows artifact or artifact family used to interpret execution, user activity, application behavior, device use, or operating-system state.
In a forensic report, run keys should be treated as a source of evidence and uncertainty, not as a shortcut to intent. The useful question is what the record supports, what it does not support, and what another source says when asked the same question.

===== Background =====
Windows leaves a large evidence trail, but the trail was not designed to be polite. Some records are loud, some are stale, and some look guilty because the operating system has terrible social skills.
This is where careful notes matter, because the artifact will not come back later to explain what the examiner forgot to write down. A Windows trace can look guilty and still prove less than the report wants it to prove.

===== Technical Description =====
Analysis usually compares artifact records with user SIDs, file paths, timestamps, execution context, registry state, event logs, and collection notes.
If the tool normalized, decoded, or reconstructed the record, that transformation belongs in the notes.

===== Forensic Relevance =====
  * **Timeline reconstruction:** Run Keys can help place activity in sequence when the time source and collection conditions are understood.
  * **Recovery analysis:** It can explain why data was recovered, missed, corrupted, or only partially reconstructed.
  * **User activity review:** It may support account or profile context, but only when independent artifacts point the same direction.
  * **Contradiction testing:** It is useful for finding places where logs, metadata, storage behavior, or accounts disagree.
  * **Reporting decisions:** It helps decide how narrow the finding should be when run keys looks interesting but not conclusive.

===== Evidence Sources =====
^ Evidence source ^ What it may show ^ Reliability limits ^ What it cannot prove alone ^
| Registry hives | May show configuration, persistence, service paths, or user activity. | Last-write times may cover multiple values. | Do not prove which value changed alone. |
| Event logs | May corroborate service creation, task execution, or administrative activity. | Coverage depends on policy. | Do not prove intent. |
| File-system metadata | May support binary path, timestamps, and signer context. | Can be rewritten by updates or restores. | Does not prove operator identity. |
| Autorun locations | May show persistence opportunities. | Legitimate software uses them constantly. | Do not make persistence malicious by default. |

===== Interpretation Limits =====
The most common error is treating run keys as proof of motive. It may support a technical event, a sequence, or a contradiction, but motive needs stronger ground.
Another bad leap is treating absence as intent. Missing data can come from retention, configuration, collection scope, sync behavior, overwriting, media behavior, or ordinary use.
A tool result should be described as a parsed or recovered record, not as the tool's opinion about what happened. Tools surface evidence; they do not understand it.
The artifact may know a path and a time. It does not know motive.

===== Common Misinterpretations =====
  * Letting one Windows artifact carry the whole timeline.
  * Treating path and time as motive.
  * Treating run keys as proof that a specific user acted, when the artifact only supports system or account context.
  * Treating a timestamp as exact truth without checking timezone, clock drift, and source semantics.
  * Ignoring benign explanations because the suspicious explanation is easier to write.

===== Example Scenario =====
A reviewer sees registry hives in the case file and asks whether it actually supports the written conclusion.
The finding may support activity in the relevant time window, but it does not prove who caused it or why.
The careful next step is to normalize the time source, compare independent artifacts, and write the finding as support rather than proof.

===== Analysis Workflow =====
  - Define the question before opening another parser: what should Run Keys help answer?
  - Preserve the source evidence and document how the registry hives were collected.
  - Record tool versions, input paths, output paths, time settings, and errors.
  - Identify observed facts before writing any interpretation.
  - Normalize time sources and document timezone, clock drift, and collection-time effects.
  - Compare at least two independent artifact families before raising confidence.
  - Consider benign explanations, automated behavior, retention, sync, and storage-device behavior.
  - Write conclusions proportionally: observed fact first, inference second, uncertainty always visible.

===== Reporting Guidance =====
Reporting on run keys should be precise enough that another analyst can retrace the claim without inheriting the original examiner's confidence.
Avoid wording that converts possibility into intent. The report should not say a user deliberately deleted, hid, wiped, or tampered with evidence unless the evidence actually supports that conclusion.
Report-ready wording:
  * The available registry hives are consistent with activity related to run keys, but they do not independently establish motive or user intent.
  * Recovery was limited under the examined conditions; additional corroboration would be required before concluding deliberate destruction.
  * The finding should be read with the collection scope, time-source limits, and alternate explanations described in this report.

===== Confidence and Reliability =====
^ Confidence level ^ What it looks like for this topic ^ How to report it ^
| Low | Registry hives exist, but collection scope, time source, or surrounding context is limited. | State the observation and keep interpretation narrow. |
| Moderate | Registry hives align with Event logs, but attribution or intent remains unresolved. | Say the artifacts support the finding, not that they prove it. |
| High | Multiple independent sources agree on sequence, source system, account context, and collection conditions. | Use stronger language, but still separate observed facts from inference. |

===== Related Concepts =====
  * [[wiki:digital_forensics:windows:amcache|Amcache]] - Often appears nearby in reviews of the same system or behavior.
  * [[wiki:digital_forensics:windows:appcompatcache|AppCompatCache]] - Useful when checking whether another artifact supports the same interpretation.
  * [[wiki:digital_forensics:windows:bam_and_dam|BAM and DAM]] - Useful when checking whether another artifact supports the same interpretation.
  * [[wiki:digital_forensics:windows:bits_jobs|BITS Jobs]] - Compare this concept when testing sequence and corroboration.
  * [[wiki:digital_forensics:windows:cloud_clipboard_artifacts|Cloud Clipboard Artifacts]] - Helps keep the finding grounded in a wider evidence pattern.
  * [[wiki:digital_forensics:windows:iconcache|IconCache]] - Compare this concept when testing sequence and corroboration.

===== Tools =====
  * [[https://ericzimmerman.github.io/#!index.md|Eric Zimmerman's Tools]] - Windows artifact parsers including MFTECmd, PECmd, EvtxECmd, and others.
  * [[https://www.sans.org/tools/kape|KAPE]] - Targeted artifact collection and processing.
  * [[https://learn.microsoft.com/en-us/sysinternals/|Sysinternals Suite]] - Microsoft Windows internals and triage tools.
  * [[https://learn.microsoft.com/en-us/sysinternals/downloads/sysmon|Sysmon]] - High-fidelity Windows endpoint telemetry.
  * [[https://github.com/keydet89/RegRipper3.0|RegRipper]] - Registry parsing framework and plugins.
  * [[https://eventlogxp.com/|Event Log Explorer]] - Windows event log viewing and filtering.
  * [[https://learn.microsoft.com/en-us/sysinternals/downloads/autoruns|Autoruns]] - Persistence and autorun location review.
  * [[https://learn.microsoft.com/en-us/sysinternals/downloads/procmon|Process Monitor]] - Process, registry, and file activity observation.
  * [[https://github.com/Yamato-Security/hayabusa|Hayabusa]] - Windows event log hunting and timeline support.
  * [[https://github.com/WithSecureLabs/chainsaw|Chainsaw]] - Windows event log search and detection.
  * [[wiki:tools:start|Forensic Tools]] - Full tool directory for the wiki.

===== Limitations of Tools =====
Tools parse, surface, and organize evidence. They do not create conclusions.
Parser output can be affected by version differences, unsupported formats, corrupted records, timezone handling, partial collection, and storage behavior outside the tool's view.
When a tool produces a strong-looking result, validate it against another tool or source where the stakes justify it. A parser can recover a fragment; it cannot tell you whether the fragment deserves a paragraph in the report.

===== References and Further Reading =====
  * [[https://learn.microsoft.com/en-us/sysinternals/|Microsoft Sysinternals]] - Windows internals and triage tools.
  * [[https://learn.microsoft.com/en-us/windows/win32/eventlog/event-logging|Microsoft Windows Event Log]] - Event logging reference.
  * [[https://www.forensicfocus.com/articles/windows-forensics-and-security/|Forensic Focus: Windows Forensics and Security]] - Windows artifact overview.
  * [[https://csrc.nist.gov/publications/detail/sp/800-86/final|NIST SP 800-86]] - Guide to integrating forensic techniques into incident response.
  * [[https://www.nist.gov/itl/ssd/software-quality-group/computer-forensics-tool-testing-program-cftt|NIST CFTT]] - Computer Forensics Tool Testing program.

===== See Also =====
  * [[wiki:digital_forensics:windows:start|Windows]]
  * [[wiki:digital_forensics:windows:amcache|Amcache]]
  * [[wiki:digital_forensics:windows:appcompatcache|AppCompatCache]]
  * [[wiki:digital_forensics:windows:bam_and_dam|BAM and DAM]]
  * [[wiki:digital_forensics:windows:bits_jobs|BITS Jobs]]
  * [[wiki:digital_forensics:windows:cloud_clipboard_artifacts|Cloud Clipboard Artifacts]]
  * [[wiki:digital_forensics:windows:iconcache|IconCache]]
  * [[wiki:digital_forensics:windows:jump_lists|Jump Lists]]
  * [[wiki:digital_forensics:windows:lnk_files|LNK Files]]
  * [[wiki:digital_forensics:windows:mounteddevices|MountedDevices]]
  * [[wiki:digital_forensics:windows:ntfs_change_journal|NTFS Change Journal]]
  * [[wiki:tools:start|Forensic Tools]]
  * [[wiki:resources:start|Project Resources]]

===== Reader Takeaway =====
Run Keys are useful when it helps explain what the evidence can support and where the limits begin. Treat run keys as one part of a corroborated record, not a shortcut to intent. The artifact may know a path and a time. It does not know motive.

===== Use Notes =====
This article is for defensive education and technical reference. It should not be treated as legal, forensic, investigative, compliance, or operational advice without qualified professional judgment.