Scripted recursive copy sweeping multiple document types

Siemphony@siemphonymediumT1119unverified

Authors a single-event stand-in for AN0531's core signal — "search for and copy information fitting set criteria such as file type" — as a PowerShell Copy-Item invocation that is both recursive and names several unrelated document extensions in one command line, which is what a scripted bulk collection sweep looks like at the point it runs versus an ordinary one-off copy. MITRE's ProcessCountThreshold and ClipboardCheckRate knobs describe a rate over repeated events, which Sigma's lack of aggregation support (no `count() by`) makes impossible to express directly, so this substitutes "several file types requested together" for "many files copied over time" as the single-event proxy for automation. Three limits are structural, and none of them is a false positive. Sysmon EventID 1 records the process command line only, so this fires solely on a sweep typed inline (powershell.exe -Command "Copy-Item -Recurse ... -Include *.doc,*.pdf"); a launch of powershell.exe -File collect.ps1, an -EncodedCommand payload, a pipeline where -Recurse and -Include sit on Get-ChildItem and the trailing Copy-Item carries neither, or plain aliasing (cpi, gci) all put none of the matched tokens on the recorded command line, because the cmdlet text lives inside the script body this event never captures. Inline invocation is one style of scripted collection and not the dominant one for real automation, so treat a silent result as unmeasured rather than clean. Second, 'recurse' is matched as a bare substring, so -Recurse:$false — recursion explicitly turned off, no tree walked — still satisfies that gate. Third, robocopy and xcopy are left out of the selector rather than OR-ed in on a different CommandLine shape, keeping the rule to one condition all four backends render. This is deliberately not the shape already published elsewhere in the corpus: T1005 matches a find/ grep search for credential and SSH-key filenames, and this matches a PowerShell copy of general document types — different tool class, different target, no overlap. UNVERIFIED AS AUTHORED — derived from MITRE ATT&CK DET0186, not hand-written and not tested by its author. Any lab result is on this rule's own page.Full description

The 4 SIEM queries below are previews produced by Siemphony’s own translator, not pySigma, and none has been executed against a real backend — review before you deploy.

detection.yml

Sentinel · KQL

Run this as a search.

DeviceProcessEvents| where ((((FolderPath endswith "\\powershell.exe" or FolderPath endswith "\\pwsh.exe") and ProcessCommandLine contains "copy-item") and ProcessCommandLine contains "recurse") and (ProcessCommandLine contains ".doc" or ProcessCommandLine contains ".xls" or ProcessCommandLine contains ".ppt" or ProcessCommandLine contains ".pdf" or ProcessCommandLine contains ".csv" or ProcessCommandLine contains ".txt"))

Splunk · SPL

Run this as a search.

index=* ((((Image="*\\powershell.exe" OR Image="*\\pwsh.exe") AND CommandLine="*copy-item*") AND CommandLine="*recurse*") AND (CommandLine="*.doc*" OR CommandLine="*.xls*" OR CommandLine="*.ppt*" OR CommandLine="*.pdf*" OR CommandLine="*.csv*" OR CommandLine="*.txt*"))

Elastic · ES|QL

Run this as a search.

FROM logs-*| WHERE ((((TO_LOWER(process.executable) LIKE "*\\\\powershell.exe" OR TO_LOWER(process.executable) LIKE "*\\\\pwsh.exe") AND TO_LOWER(process.command_line) LIKE "*copy-item*") AND TO_LOWER(process.command_line) LIKE "*recurse*") AND (TO_LOWER(process.command_line) LIKE "*.doc*" OR TO_LOWER(process.command_line) LIKE "*.xls*" OR TO_LOWER(process.command_line) LIKE "*.ppt*" OR TO_LOWER(process.command_line) LIKE "*.pdf*" OR TO_LOWER(process.command_line) LIKE "*.csv*" OR TO_LOWER(process.command_line) LIKE "*.txt*"))

Wazuh · XML rule

Deploy to your manager — this is a rule, not a search.

<group name="sigma,windows,process_creation,">  <!-- Rule ids must be unique on your manager; 100000+ is the user range. -->  <rule id="100000" level="7">    <!-- Set <if_sid> to the decoder/base rule for windows so this only evaluates relevant events. -->    <field name="Image" type="pcre2">(?i)(\\powershell\.exe$|\\pwsh\.exe$)</field>    <field name="CommandLine" type="pcre2">(?i)copy-item</field>    <field name="CommandLine" type="pcre2">(?i)recurse</field>    <field name="CommandLine" type="pcre2">(?i)(\.doc|\.xls|\.ppt|\.pdf|\.csv|\.txt)</field>    <description>Scripted recursive copy sweeping multiple document types</description>    <mitre>      <id>T1119</id>    </mitre>  </rule></group>

At a glance

Logsource: process_creation / windows
Level: medium
Category: data_staging_exfil
Sigma status: experimental
Backends: 4 of 4 rendered
Updated: 30 Aug 2026
Rule ID: aad6ea77-2853-4827-abf9-9c877c56e2ab

ATT&CKT1119

T1119Automated Collection

ATT&CK® is a registered trademark of The MITRE Corporation. These mappings are the rule author’s, and are neither reviewed nor endorsed by MITRE.

Lessondata staging exfil

data staging exfil

01 · What this rule detects

Collection before departure: files gathered from profiles and shares, compressed — often into a password-protected archive — dropped into a staging directory, and then moved out of the environment. The staging half is the reliably detectable half, and it usually happens close enough to the transfer that catching it still leaves time to act.

02 · Telemetry you need

Process creation with full command lines for archiving utilities — 7z, rar, WinRAR's console build, tar.exe, makecab, and PowerShell Compress-Archive; the switches are the interesting part, so a truncated command line is worthless here
Sysmon EventID 11 (FileCreate) for archive extensions written under Temp, ProgramData, Public and the recycle bin
Sysmon EventID 3 (NetworkConnect) with DestinationIp, DestinationHostname, DestinationPort and Initiated, tied to the uploading process
Proxy, firewall or flow logs with bytes sent per session — these are the only records that carry volume and direction, and neither Sysmon nor the Security log has them
Security 5145 (detailed file share access) and 4663 (object access) for reads across shares — 5145 requires Audit Detailed File Share, 4663 requires Audit Object Access plus a SACL on the data itself, and all of it is off by default
DNS query logs, for the cloud storage and tunnelling destinations the transfer half resolves

03 · How the logic works

Split this into two rules, because it is really two behaviours with different telemetry. The staging rule is a process rule: an archiving utility invoked with a recursion switch, an output path under a staging directory, or a password flag, over a source path that holds user data or share content. Anchor on the switches and the paths rather than the tool name — a renamed rar.exe still needs the same arguments to do the same job — and keep the password flag as its own higher-severity condition, since ordinary backup rarely needs one. The transfer rule is a volume and destination rule, and it does not express well in Sigma: there is no aggregation, so the bytes-sent threshold and the per-destination grouping live in the backend query or the correlation layer. The pairing that makes both worth having is sequence — an archive created, then that same file opened by a process that immediately connects outbound.

04 · Common false positives

Backup and archiving agents doing precisely this by design, including Veeam, Acronis and Windows Server Backup
Sanctioned cloud sync clients — OneDrive, Dropbox, Box — moving large volumes to destinations that are supposed to receive them
Developers and build agents compressing source trees, artefacts and logs, and CI runners uploading them
IT collecting diagnostic bundles and log archives for a vendor support case
E-discovery, DLP and data-classification scanners reading broadly across shares, and migration tooling moving mailboxes or file servers

05 · Detection maturity

Tier 1 — starting point: Matches archiving utility command lines by tool name.
Tier 2 — tuned: Matches recursion and password switches over sensitive source paths, with backup and sync clients excluded by signed path.
Tier 3 — correlated: Correlates archive creation with the subsequent read and outbound transfer of that file, scored on destination novelty and volume.

Tuning & false positivesexperimental · 3 declared

experimental (current stage)
test
stable

Written and reasoned about, but not yet run against real telemetry. Expect to tune it before trusting it.

Tuning tips

Password-protected archives of user data are the highest-payoff single condition in this category; alert on those before anything else
Scope the staging rule by source path — profile directories and file shares — rather than by the utility, so unfamiliar tools still match
Exclude sanctioned cloud sync by the signed path of the client process, not by destination domain, or you excuse anything that can reach the same domain
Keep staging and transfer as separate alerts so each stays actionable, and correlate them for severity rather than merging them into one rule
Where the archive is written matters more than what wrote it: a user-writable staging directory holding an archive of someone else's data is the shape you are looking for

Declared by the author

User-profile migration and onboarding/offboarding scripts run by helpdesk tooling, which use Copy-Item -Recurse against a user's Documents folder filtered to common office extensions as their normal operating pattern — in an environment with active migration or provisioning waves this is the largest source of matches.
Backup, e-discovery or legal-hold agents that periodically sweep user document folders and copy matching file types to a review or backup repository as part of routine, scheduled compliance operations.
Software packaging and imaging scripts (SCCM/Intune build tasks) that stage a fixed set of document templates into a reference image using the same cmdlet and flag combination.

Noise check0 of 30 fired

Fires on none of the 30 ordinary behaviours the baseline holds for this log type. That is not a precision guarantee: it means only not caught by these 30 behaviours, and the next ordinary behaviour nobody wrote down may still match.

Run live against 65 hand-written events of ordinary corporate activity — backup agents, software distribution, patching, helpdesk scripts, EDR. It says nothing about how often the rule would fire in production, which is the number that actually exhausts an analyst.

Verification1 run

Not verified

Siemphony benign baseline
completed
Siemphony IR · 3 Sept 2026
No events matched, so this run proves nothing either way.

Test it yourselfin your browser

Runs here in your browser against public attack recordings. A preview — not the verification badge.

Recordings of T1119, evaluated against their process_creation events.

Share this rule

/u/siemphony/aad6ea77-2853-4827-abf9-9c877c56e2ab

More from this repertoire

All compositions

Verdicts · reactions · comments

Community

Verdicts from engineers who actually deployed it, and the conversation around it.

Not yet reported on

Nobody has run this in a real environment and said what happened.

Verdicts from engineers who deployed it

0 cast

No verdicts reported yet. The first one is worth more than the tenth — say what you ran it against.

No comments yet. Someone who deploys this will have something to say about it.

Scripted recursive copy sweeping multiple document types

The detection

detection.yml

Sentinel · KQL

Splunk · SPL

Elastic · ES|QL

Wazuh · XML rule

Community

Verdicts from engineers who deployed it