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.
====== Service Persistence ======

===== Definition =====
Service Persistence is a malware or detection concept used to interpret code behavior, suspicious execution, defensive telemetry, or incident response findings.
In a forensic report, service persistence 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 =====
Malware analysis is where tools can be brilliant and still not know what the evidence means. A rule hit is a lead, not a courtroom verdict with better syntax highlighting.
A reviewer should be able to follow the claim from source evidence to cautious conclusion without spelunking through unsupported confidence. The sandbox can watch malware perform. It still cannot swear the production host saw the same show.

===== Technical Description =====
Review compares static properties, dynamic behavior, memory evidence, process lineage, persistence records, network traffic, and detection-rule context.
If the tool normalized, decoded, or reconstructed the record, that transformation belongs in the notes.

===== Forensic Relevance =====
  * **Reporting decisions:** It helps decide how narrow the finding should be when service persistence looks interesting but not conclusive.
  * **Timeline reconstruction:** Service Persistence 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.

===== Evidence Sources =====
^ Evidence source ^ What it may show ^ Reliability limits ^ What it cannot prove alone ^
| Executable or script sample | May show imports, strings, capabilities, packers, or obfuscation. | Static features can be misleading or intentionally confusing. | Do not prove deployment context. |
| Sandbox output | May show runtime behavior, files, registry changes, and network activity. | Sandbox awareness and environment mismatch can distort behavior. | Does not prove what happened on the endpoint. |
| Memory and process evidence | May show injected code, command lines, sockets, and modules. | Collection timing matters. | Does not identify the actor alone. |
| Detection rules and alerts | May support triage and hunt leads. | False positives and rule scope must be documented. | Do not replace artifact analysis. |

===== Interpretation Limits =====
The most common error is treating service persistence 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.
Packed code can be suspicious without explaining who ran it, why, or whether the endpoint ever saw the same behavior.

===== Common Misinterpretations =====
  * Ignoring false positives because the alert name sounded expensive.
  * Treating service persistence 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.
  * Treating a detection rule hit as a verdict.

===== Example Scenario =====
A responder notices executable or script sample that appears to line up with a disputed timeline.
The record may be meaningful, but it has to be compared with sandbox output before the report gets confident.
The analyst preserves the source, exports the relevant records, checks sandbox output, and documents what remains unresolved.

===== Analysis Workflow =====
  - Define the question before opening another parser: what should Service Persistence help answer?
  - Preserve the source evidence and document how the executable or script sample was 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 service persistence 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 executable or script sample is consistent with activity related to service persistence, but it does 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 | Executable or script sample exists, but collection scope, time source, or surrounding context is limited. | State the observation and keep interpretation narrow. |
| Moderate | Executable or script sample aligns with Sandbox output, 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:memory_malware_ir:persistence_mechanisms|Persistence Mechanisms]] - Compare this concept when testing sequence and corroboration.
  * [[wiki:digital_forensics:memory_malware_ir:scheduled_task_persistence|Scheduled Task Persistence]] - Helps keep the finding grounded in a wider evidence pattern.
  * [[wiki:digital_forensics:memory_malware_ir:anti_debugging|Anti-Debugging]] - Useful when checking whether another artifact supports the same interpretation.
  * [[wiki:digital_forensics:memory_malware_ir:anti_vm_checks|Anti-VM Checks]] - Often appears nearby in reviews of the same system or behavior.
  * [[wiki:digital_forensics:memory_malware_ir:api_hashing|API Hashing]] - Compare this concept when testing sequence and corroboration.
  * [[wiki:digital_forensics:memory_malware_ir:autoruns|Autoruns]] - Compare this concept when testing sequence and corroboration.

===== Tools =====
  * [[https://remnux.org/|REMnux]] - Linux malware analysis environment.
  * [[https://github.com/mandiant/flare-vm|FLARE-VM]] - Windows malware analysis toolkit.
  * [[https://ghidra-sre.org/|Ghidra]] - Reverse engineering suite.
  * [[https://virustotal.github.io/yara/|YARA]] - Pattern matching for malware triage.
  * [[https://github.com/mandiant/capa|capa]] - Capability detection for executable analysis.
  * [[https://github.com/mandiant/flare-floss|FLOSS]] - String decoding and extraction for malware analysis.
  * [[https://github.com/volatilityfoundation/volatility3|Volatility 3]] - Memory analysis and plugin-driven investigation.
  * [[https://github.com/Velocidex/WinPmem|WinPmem]] - Windows memory acquisition.
  * [[https://github.com/504ensicsLabs/LiME|LiME]] - Linux memory acquisition.
  * [[https://github.com/microsoft/avml|AVML]] - Linux memory acquisition for cloud and endpoint response.
  * [[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://volatility3.readthedocs.io/|Volatility 3 documentation]] - Memory analysis documentation.
  * [[https://yara.readthedocs.io/|YARA documentation]] - YARA rule documentation.
  * [[https://www.cisa.gov/resources-tools|CISA resources and tools]] - CISA cybersecurity resources and tools.
  * [[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:memory_malware_ir:start|Memory Malware Ir]]
  * [[wiki:digital_forensics:memory_malware_ir:persistence_mechanisms|Persistence Mechanisms]]
  * [[wiki:digital_forensics:memory_malware_ir:scheduled_task_persistence|Scheduled Task Persistence]]
  * [[wiki:digital_forensics:memory_malware_ir:anti_debugging|Anti-Debugging]]
  * [[wiki:digital_forensics:memory_malware_ir:anti_vm_checks|Anti-VM Checks]]
  * [[wiki:digital_forensics:memory_malware_ir:api_hashing|API Hashing]]
  * [[wiki:digital_forensics:memory_malware_ir:autoruns|Autoruns]]
  * [[wiki:digital_forensics:memory_malware_ir:beaconing|Beaconing]]
  * [[wiki:digital_forensics:memory_malware_ir:command_and_control_traffic|Command and Control Traffic]]
  * [[wiki:digital_forensics:memory_malware_ir:command_line_artifacts|Command Line Artifacts]]
  * [[wiki:digital_forensics:memory_malware_ir:containment_notes|Containment Notes]]
  * [[wiki:tools:start|Forensic Tools]]
  * [[wiki:resources:start|Project Resources]]

===== Reader Takeaway =====
Service Persistence is useful when it helps explain what the evidence can support and where the limits begin. Treat service persistence as one part of a corroborated record, not a shortcut to intent. The sandbox can watch malware perform. It still cannot swear the production host saw the same show.

===== 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.