Analyzing Windows Prefetch With Python

Analyzing Windows Prefetch with Python

Overview

Windows Prefetch files (.pf) record application execution data including executable names, run counts, timestamps, loaded DLLs, and accessed directories. This skill covers parsing Prefetch files using the windowsprefetch Python library to reconstruct execution timelines, detect renamed or masquerading binaries by comparing executable names with loaded resources, and identifying suspicious programs that may indicate malware execution or lateral movement.

When to Use

• When investigating security incidents that require analyzing windows prefetch with python
• When building detection rules or threat hunting queries for this domain
• When SOC analysts need structured procedures for this analysis type
• When validating security monitoring coverage for related attack techniques

Detection Gaps & Validation

• Prefetch can be disabled or absent. It is off by default on Windows Server, and EnablePrefetcher (HKLM\SYSTEM...\Memory Management\PrefetchParameters) may be set to 0 by hardening or by an attacker. No .pf does not mean no execution — confirm the setting before concluding.
• Run-time history is shallow. Windows 8.1/10/11 store only the last 8 execution timestamps; Windows 7 stores one. Earlier runs are lost. The "first executed" estimate is the .pf creation time minus ~10 seconds, not an exact start.
• Renamed/masquerading binaries: the hash in EXECUTABLE-HASH.pf is derived from the full path, so the same binary run from two paths yields two different .pf files — and a renamed malware keeps its real loaded-DLL fingerprint. Compare referenced DLLs/directories, not just the name.
• Anti-forensics that defeat this analysis: deletion of C:\Windows\Prefetch\*.pf, disabling the Prefetcher, and .pf timestomping. Carve deleted .pf (header MAM\x04 / SCCA) from unallocated space and VSS.
• Validate / cross-corroborate: confirm execution against Amcache, ShimCache/AppCompatCache, SRUM SRUDB.dat (app usage + bytes), and Security.evtx EID 4688. The volume serial in the .pf should match the system under analysis.
• Interpretation false positives: a high run count on svchost.exe/explorer.exe is normal; flag on path, first-seen during incident window, and loaded-resource anomalies, never run count alone.

Prerequisites

• Python 3.9+ with windowsprefetch library (pip install windowsprefetch)
• Windows Prefetch files from C:\Windows\Prefetch\ (versions 17-30 supported)
• Understanding of Windows Prefetch file naming conventions (EXECUTABLE-HASH.pf)

Steps

Step 1: Collect Prefetch Files

Gather .pf files from target system's C:\Windows\Prefetch\ directory.

Step 2: Parse Execution History

Extract executable name, run count, last execution timestamps, and volume information.

Step 3: Detect Suspicious Execution

Flag known attack tools (mimikatz, psexec, etc.), renamed binaries, and unusual execution patterns.

Step 4: Build Execution Timeline

Reconstruct chronological execution timeline from all Prefetch files.

Expected Output

JSON report with execution history, suspicious executables, renamed binary indicators, and timeline reconstruction.

Example Output

$ python3 prefetch_analyzer.py --dir /evidence/Windows/Prefetch --output /analysis/prefetch_report

Windows Prefetch Analyzer v2.1
================================
Source: /evidence/Windows/Prefetch/
Prefetch Format: Windows 10 (MAM compressed, version 30)
Files Found: 234

--- Execution Timeline (Incident Window: 2024-01-15 to 2024-01-18) ---
Last Executed (UTC)     | Run Count | Filename                    | Hash     | Path
------------------------|-----------|-----------------------------|----------|------------------------------------------
2024-01-15 14:33:15     | 1         | Q4_REPORT.XLSM-2A1B3C4D.pf | 2A1B3C4D | C:\Users\jsmith\Downloads\Q4_Report.xlsm
2024-01-15 14:35:44     | 1         | POWERSHELL.EXE-A2B3C4D5.pf  | A2B3C4D5 | C:\Windows\System32\WindowsPowerShell\v1.0\powershell.exe
2024-01-15 14:36:30     | 3         | UPDATE_CLIENT.EXE-B3C4D5E6.pf| B3C4D5E6| C:\ProgramData\Updates\update_client.exe
2024-01-15 15:10:22     | 1         | NETSCAN.EXE-C4D5E6F7.pf     | C4D5E6F7 | C:\Users\jsmith\Downloads\netscan.exe
2024-01-16 02:28:00     | 1         | PROCDUMP64.EXE-D5E6F7A8.pf  | D5E6F7A8 | C:\Windows\Temp\procdump64.exe
2024-01-16 02:30:15     | 2         | MIMIKATZ.EXE-E6F7A8B9.pf    | E6F7A8B9 | C:\Windows\Temp\mimikatz.exe
2024-01-16 02:40:00     | 4         | PSEXEC.EXE-F7A8B9C0.pf      | F7A8B9C0 | C:\Users\jsmith\AppData\Local\Temp\psexec.exe
2024-01-17 02:45:00     | 1         | SDELETE64.EXE-A8B9C0D1.pf   | A8B9C0D1 | C:\Windows\Temp\sdelete64.exe
2024-01-18 03:00:45     | 1         | WEVTUTIL.EXE-B9C0D1E2.pf    | B9C0D1E2 | C:\Windows\System32\wevtutil.exe

--- Renamed Binary Detection ---
ALERT: UPDATE_CLIENT.EXE loaded DLLs consistent with Cobalt Strike beacon:
  Referenced DLLs: wininet.dll, ws2_32.dll, advapi32.dll, dnsapi.dll, netapi32.dll
  Volume: \VOLUME{01d94f2a3b5c7d8e-A4E73F21} (C:)
  Directories referenced:
    C:\ProgramData\Updates\
    C:\Windows\System32\

--- Execution Frequency Analysis ---
Most Executed (Top 5):
  1. SVCHOST.EXE          (267 runs)
  2. CHROME.EXE           (189 runs)
  3. EXPLORER.EXE         (156 runs)
  4. RUNTIMEBROKER.EXE    (134 runs)
  5. OUTLOOK.EXE          (98 runs)

First-Time Executions (Never seen before incident window):
  6 executables first run between 2024-01-15 and 2024-01-18

Summary:
  Total prefetch files:         234
  Suspicious executables:       6
  Renamed binary indicators:    1 (update_client.exe)
  Anti-forensics tools:         2 (sdelete64.exe, wevtutil.exe)
  JSON report: /analysis/prefetch_report/prefetch_timeline.json