Join our Newsletter — 33% off our NHI Course
Home› Guides› T1560 and T1074 Archive and Stage Collected Data…
Guide Attack Techniques & Malware

T1560 and T1074 Archive and Stage Collected Data Defensive Guide: Spotting Data Packed for Theft

PDF for newsletter subscribers · free to read online

← All guides
By NHI Mgmt Group, edited by Lalit Choda Updated 27 October 2026 11 min read
On this page

Before stolen data leaves a network it usually sits, gathered and compressed, in a temporary folder, a public directory, a file server or a cloud stage, and that pause is one of the best chances defenders get to stop the theft. MITRE ATT&CK describes the two behaviours as T1560 Archive Collected Data (compressing or encrypting collected data before exfiltration) and T1074 Data Staged (gathering it in a central location first). Both sit in the Collection tactic, both rely heavily on legitimate tools such as 7-Zip, WinRAR, tar and makecab, and in ATT&CK v20 both have few preventive mitigations, which makes detection the main control. This guide covers how archiving and staging work on endpoints, servers, hypervisors and cloud data platforms, how to detect them, how to make them harder, and what recent intrusions teach. It is for SOC analysts, detection engineers, data security and cloud teams. For exfiltration channels themselves, see the Data Exfiltration Guide.

Key takeaways

  • Staging is the last quiet step before loss. Alerts that fire on archive creation or data aggregation buy time that alerts on exfiltration do not.
  • Legitimate tools do the work. ATT&CK lists tar, zip, diantz, makecab, xcopy, 7-Zip, WinRAR and WinZip for T1560.001; context, not the binary, makes it malicious.
  • Passwords and renames are tells. Password-protected or multi-volume archives, and archive tools running under new names, are rare in normal administration.
  • Cloud staging looks like SQL and API calls. UNC5537 staged Snowflake data with CREATE TEMPORARY STAGE and COPY INTO before pulling it with GET, so data platform audit logs are part of the detection surface.
  • Mitigations are thin. v20 maps only M1047 Audit to T1560 and none to T1074, so build controls from DLP, least privilege on data, application control and good logging.

How the attack works

ATT&CK describes both techniques as preparation for exfiltration. Compression shrinks what crosses the network and helps obscure it; encryption hides the content from inspection. Staging in one place lets the intruder make fewer connections to their C2 server and so evade detection.

TechniqueWhat it coversPlatforms in v20Examples recorded in ATT&CK v20
T1560.001 Archive via UtilityBuilt-in or third-party archivers: tar, zip, diantz and makecab (.cab), xcopy, 7-Zip, WinRAR, WinZipWindows, Linux, macOSHAFNIUM using 7-Zip and WinRAR; Operation CuckooBees using makecab; Volt Typhoon archiving ntds.dit as a multi-volume password-protected 7-Zip archive; Sea Turtle using tar on email data
T1560.002 Archive via LibraryCompression linked into code (Python rarfile, libzip, zlib, bzip2) rather than a spawned utilityWindows, Linux, macOSLazarus Group malware compressing gathered data with zlib; ShinyHunters compressing collected data
T1560.003 Archive via Custom MethodXOR, stream ciphers or home-grown compression with no external libraryWindows, Linux, macOSMustang Panda and Kimsuky using RC4; UNC3886 XOR-encrypting and gzip-compressing captured credentials
T1074.001 Local Data StagingCollecting files in one local directory, local database or even the RegistryWindows, Linux, macOS, ESXiSidewinder using a temporary folder; Operation Wocao staging archives in a temporary directory; Storm-1811 staging captured credentials locally
T1074.002 Remote Data StagingAggregating data from many systems onto one internal system or cloud instanceWindows, Linux, macOS, ESXi, IaaSAPT28 staging archives on an Outlook Web Access server; Sea Turtle staging email archives in a public web directory; menuPass staging on MSP systems; APT29 staging password-protected archives during the SolarWinds compromise

The surrounding chain is predictable: collection from local systems, shares or repositories (T1005, T1039, T1213), often automated (T1119), then staging and archiving, then exfiltration over the C2 channel (T1041) or to cloud storage (T1567.002). In cloud environments ATT&CK notes adversaries may create a cloud instance (T1578.002) purely to stage data. See the ATT&CK Collection Tactic Guide.

Where it shows up

  • Domain controllers: CISA's February 2024 Volt Typhoon advisory (AA24-038A) describes actors using WMIC to stage extracted NTDS.dit and registry files in C:\Users\Public\pro and C:\Windows\Temp\tmp, then compressing them with a renamed archive utility, likely rar.exe running as ronf.exe. ATT&CK separately records Volt Typhoon using a password-protected multi-volume 7-Zip archive for ntds.dit. See the Volt Typhoon Threat Actor Profile.
  • Web-facing servers: staging in a web directory (Sea Turtle) or on OWA (APT28) lets the data be downloaded over ordinary HTTPS.
  • Workstations of administrators: CISA notes a History.zip file, likely containing Chrome user data, staged in a network administrator's Downloads folder.
  • File servers and ransomware operations: double-extortion groups such as INC Ransom and Play stage and compress data before encryption.
  • Cloud data platforms: Mandiant reported on 10 June 2024 that UNC5537 created temporary stages in victims' Snowflake accounts, ran COPY INTO with gzip compression and 5 GB file sizes, then used GET to pull the files. Temporary stages are deleted when the session ends, so evidence lives in query history, not in storage.
  • Hypervisors and IaaS: v20 analytics cover ESXi snapshots and VMFS writes, and CloudTrail GetObject and CopyObject patterns that aggregate data across buckets.

Impact

  • Data theft at scale: archiving turns thousands of files or whole tables into a few transferable objects.
  • Credential compromise: archived ntds.dit and SYSTEM hives give the intruder every domain password hash (T1003.003 NTDS), forcing a full credential reset.
  • Blind DLP: encrypted or password-protected archives defeat content inspection at the boundary, so staging-time detection is the only chance to see the content.
  • Extortion leverage: staged data is what ransomware groups publish when payment is refused.

Detection

ATT&CK v20's analytics correlate archiver execution with creation of large compressed or encrypted files in temporary or staging directories (Security Event ID 4688, Sysmon 1, 7 and 11 on Windows; auditd execve and file creation in /tmp or /var/tmp on Linux; zip, ditto and hdiutil started by non-terminal parents on macOS), plus object-access auditing (4656, 4663, 4670) and SMB client events for remote staging.

SignalTelemetryTechniqueTuning note
7z, rar, makecab, diantz, tar or Compress-Archive run by a service account, on a server, or with password (-p) or volume-split options4688 or Sysmon 1 command lines, EDRT1560.001Baseline backup and software deployment jobs; password and split options are rare elsewhere
Archiver binary whose original file name differs from the image nameSysmon 1 OriginalFileName, EDR file metadataT1560.001 with T1036.003Catches the renamed rar.exe pattern from the Volt Typhoon advisory
New .zip, .rar, .7z, .cab or multi-part archives over a size threshold in C:\Users\Public, C:\Windows\Temp, C:\ProgramData, web roots, /tmp or /var/tmpSysmon 11, auditd, EDR file eventsT1074.001Directory list plus size gate keeps volume manageable
NTDS.dit copied or mounted outside its normal pathESENT application log events 216, 325, 326, 327 and 637 (CISA AA24-038A)T1074.001 with T1003.003CISA highlights event 325, a new NTDS.dit in a non-standard directory
One host pulling many files from many shares in a short window (robocopy, copy-item from UNC paths, scp, rsync)SMB client events, 5145 share access, flow logsT1074.002Compare with the host's role; file servers and backup hosts excepted
CREATE TEMPORARY STAGE followed by large COPY INTO and GET; unusual client application IDsSnowflake query and login historyT1074 (cloud)NHI Mgmt Group's Snowflake analysis recommends alerting on large COPY or GET operations to temporary stages
Process with high read volume followed by a single large file writeEDR file I/O telemetryT1560.002, .003Library and custom archiving spawn no archiver, so watch the I/O pattern

Measuring it

  • Coverage: share of servers and domain controllers sending process command lines and file creation events to the SIEM.
  • Staging dwell time: in exercises, minutes from archive creation to alert; the aim is to alert before any exfiltration test completes.
  • Unapproved archivers: number of hosts with archive utilities outside the approved list (the M1047 Audit check), trending down.
  • Data platform logging: percentage of cloud data warehouses and buckets with query or data event logging retained for at least a year.
  • False positive rate: archive alerts closed as benign per week, tracked to keep the rule tuned rather than disabled.

Prevention and hardening

ATT&CK v20 lists one mitigation for T1560, M1047 Audit (scan systems for unauthorised archival utilities), and none for T1074. The following controls reduce the opportunity and raise the noise an intruder must make.

  1. Control archivers on servers: use application control to allow 7-Zip, WinRAR and similar tools only where a documented job needs them, and remove them from domain controllers (NIST SP 800-53 CM-7 Least Functionality).
  2. Lock down common staging paths: restrict write access to C:\Users\Public, web roots and shared temp directories on servers (ATT&CK M1022 Restrict File and Directory Permissions), and monitor what remains.
  3. Reduce what can be collected: least privilege on shares, repositories and data warehouses (AC-6) limits the size of any staged set. Service accounts used by integrations should not have bulk read across all data.
  4. Restrict cloud export paths: in data platforms, limit who can create stages and run unload commands, and prefer network policies or allowlists for client access. Mandiant's UNC5537 report adds MFA on every account and credential rotation.
  5. Apply DLP where content is visible: M1057 Data Loss Prevention on endpoints and sanctioned SaaS sees files before they are archived and encrypted; boundary DLP cannot see inside password-protected archives, so flag those by type.
  6. Log for the investigation you will need: enable command-line auditing, Sysmon or EDR file events, object-access auditing on sensitive shares, and data platform query history (AU-12 Audit Record Generation).

Rolling it out

  1. Inventory legitimate archive use for four weeks (backup agents, deployment tools, log rotation) and record owners.
  2. Deploy the archive and staging detections in alert-only mode on domain controllers and crown-jewel servers first.
  3. Move archivers to an allowlist on servers in audit mode, then enforce by server tier; keep the previous policy version for rollback.
  4. Extend to workstations and cloud data platforms, then run an authorised staging exercise to confirm the alerts fire.

Choosing an approach: small teams should start with three rules: archivers with password or split options, archives in public and temp paths, and NTDS.dit outside its path. Large organisations should add data access analytics on file servers, cloud data platform query monitoring and DLP integration, and tie alerts to automatic host isolation for domain controllers.

Lessons from incidents

  • Cloud staging leaves query trails, not files. In the Snowflake breach of 2024, attackers created temporary stages and copied whole tables out. Temporary stages vanish at session end, so only teams with query history monitoring could see the staging step.
  • Living-off-the-land archiving needs behavioural rules. The Volt Typhoon advisory shows a renamed archiver, WMIC and built-in paths: no malware signature would have fired, but rules on archive locations, renamed binaries and ESENT events could.

Where identity fits

The data most often archived in espionage cases is credential material: ntds.dit and registry hives, browser profile data and captured credentials (UNC3886 and Storm-1811 in ATT&CK). On data platforms, the staging commands run under a valid user or service identity. Treat any confirmed staging of credential stores as a full credential compromise, rotate service account and integration credentials, and review which non-human identities hold bulk read rights. See the MITRE ATT&CK T1078 Valid Accounts Guide.

Framework mapping

FrameworkReferenceRelevance
ATT&CK v20 mitigationsM1047 Audit; M1022 Restrict File and Directory Permissions; M1057 Data Loss PreventionFind unapproved archivers, restrict staging paths, inspect content before encryption
MITRE D3FENDD3-FCA File Creation Analysis; D3-FAPA File Access Pattern Analysis; D3-PA Process Analysis; D3-PHDURA Per Host Download-Upload Ratio AnalysisNew archives in odd paths, bulk read patterns, archiver process context, the upload that follows
NIST SP 800-53 Rev. 5CM-7; AC-6; AU-12; SI-4(18) Analyze Traffic and Covert Exfiltration; SC-7(10) Prevent ExfiltrationLeast functionality and privilege, audit generation, exfiltration analytics
NIST CSF 2.0PR.DS-01; DE.CM-03Protect data at rest, monitor personnel and technology usage

Practitioner checklist

  • Inventory and allowlist archive utilities on servers; remove them from domain controllers.
  • Alert on archivers with password or volume-split options outside approved jobs.
  • Alert on archivers whose original file name differs from the image name.
  • Watch C:\Users\Public, C:\Windows\Temp, web roots, /tmp and /var/tmp for large new archives.
  • Collect ESENT events 216, 325, 326, 327 and 637 from domain controllers.
  • Monitor hosts that pull from many shares in a short window.
  • Enable and retain data platform query history; alert on temporary stages with large COPY or GET.
  • Restrict who can create stages and unload data in cloud warehouses.
  • Run an authorised staging exercise each quarter and measure time to alert.

Standards and references

Related NHI Mgmt Group resources: Data Exfiltration Guide · Data Exfiltration Playbook · ATT&CK Collection Tactic Guide · T1102 Web Service Defensive Guide · MITRE ATT&CK Guide for Defenders

PDF for subscribers

Take this guide with you

Every section, table and checklist in one PDF, with links to the sources and related guides. Subscribe to our weekly NHI & AI Identity Journal and we’ll email you a download link.

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

NHIMG Editorial Note
Researched and written by the NHI Mgmt Group editorial team using AI tools, under the editorial direction of Lalit Choda. Facts are drawn from the sources listed on this page and checked by our automated quality process. Last updated 27 October 2026. Spotted an error? Use our contact page.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM and Agentic AI security. nhimg.org

PDF Downloads for Newsletter Subscribers

Get the guide

Enter your details and we’ll email you a download link.

The full guide is free to read online. The PDF is a thank you for subscribing.

Fill in all fields marked * and tick the PDF box to continue.