Last seven days
- First activity
- Aug 18, 2026
- Last activity
- Aug 23, 2026
- Feed role
- C2
- Host form
- 2 IP / 0 hostnames
PoshC2 is a Python-based post-exploitation command-and-control framework with multiple agent formats and cross-platform support.
Profile source: Mallory opens in a new tabPoshC2
PoshC2 is a Python-based post-exploitation command-and-control framework with multiple agent formats and cross-platform support. It is used as an offensive implant and operator framework for maintaining access, executing commands, and extending post-compromise activity on victim systems. It is publicly known as a dual-use framework and has been observed in real intrusions, including activity associated with APT33.
PoshC2 supports command-and-control over HTTP and HTTPS and provides operators with modules for host and domain discovery, credential access, privilege escalation, and persistence. Documented capabilities include keystroke logging; capturing credentials through spoofed Outlook authentication prompts; searching local and remote files for passwords; decrypting credentials stored in RDCMan configuration data; enumerating local and domain user account information; enumerating network adapter information; and executing tasks through WMI. It also includes multiple methods to bypass Windows User Account Control and has GetSystem-style functionality implemented through PowerShell to obtain SYSTEM-level execution.
On Windows, PoshC2 has been documented using PowerShell launchers, registry-based payload storage, Run key persistence, scheduled tasks, and WMI event subscription persistence. It can also load external offensive PowerShell tooling, expanding its credential theft and post-exploitation functionality. Operational reporting has shown PoshC2 delivered through malicious HTA and PowerShell chains, including employment-themed lure activity attributed to APT33, and it has also been deployed as a follow-on payload by other malware such as SystemBC. PoshC2 is most accurately characterized as a post-exploitation framework and backdoor used for interactive intrusion operations rather than a commodity self-propagating malware family.
C2 tracking
Derp observations, rolling seven-day window
Reported operators
PoshC2 (specifically as used by APT33)
Exploited software
MITRE ATT&CK
Reporting
Microsoft disclosed CVE-2026-62781, a high-severity remote code execution flaw in the Windows RPC Runtime Library that allows an unauthenticated attacker to execute code over the network. The heap-based buffer overflow, tracked as CWE-122 and scored CVSS 8.1, affects a broad range of Windows client and server platforms, including Windows 10, Windows 11, and Windows Server 2012, 2016, 2019, 2022, and 2025, according to the CVE entry and Microsoft advisory. The exposure is notable because exploitation of remote services has repeatedly enabled rapid lateral movement and internal propagation across Windows environments. MITRE ATT&CK maps this activity to T1210, citing past abuse of SMB, Netlogon, RPC, and Print Spooler flaws by threats such as WannaCry, NotPetya, TrickBot, Conficker, and multiple intrusion groups; Splunk has also tied remote-service exploitation detections to the same technique in prior RCE-related content. Security teams are likely to treat the new RPC flaw as a priority patching issue given its network-reachable nature and the history of Windows remote-service vulnerabilities being used for enterprise-wide spread.
Brazilian educational institutions faced a sustained wave of cyber incidents in incident-response cases reviewed from January 2025 through June 2026, with attackers most often gaining access through valid accounts, exploitation of public-facing applications, insider activity, and weak patch management. High-severity cases were dominated by ransomware, particularly DragonForce and LockBit 3, and private institutions were more frequently affected than public ones. Investigators also found that outdated and unpatched systems, including Windows 10 deployments kept past end of support and unpatched Windows Server 2016 hosts, materially increased exposure. Representative intrusions showed attackers relying on common but effective techniques rather than novel tradecraft. One LockBit case involved a custom deployment built from the leaked builder and spread with PsExec, aligning with the well-documented abuse of Windows service execution for lateral movement and payload launch. Another DragonForce intrusion used AnyDesk and log wiping to maintain access and hinder response, while a separate insider case involved a Python keylogger installed on a shared machine to capture credentials. The findings underscore the need for MFA, least privilege, removal of shared accounts, tighter control of remote-access tools, stronger backups, centralized logging, longer EDR retention, and faster patching.
Splunk published a Windows endpoint analytic that detects suspicious child processes launched by consent.exe, a behavior strongly associated with User Account Control (UAC) bypass and privilege escalation. Because consent.exe normally displays the UAC elevation prompt rather than spawning executables, the detection treats such process creation as anomalous, excluding WerFault.exe as a known crash-related exception. The analytic maps to MITRE ATT&CK techniques T1548.002, T1068, and T1059, and is designed for telemetry from Sysmon, Windows Security Event ID 4688, and CrowdStrike ProcessRollup2 data normalized into Splunk's Endpoint data model. The release aligns with MITRE ATT&CK documentation showing UAC bypass remains a widely used post-compromise technique across ransomware operators, commodity malware, and state-linked intrusion groups. ATT&CK lists methods including COM abuse through CMSTPLUA, scheduled task abuse such as SilentCleanup, registry hijacks involving ms-settings and mscfile, and abuse of trusted Windows binaries including fodhelper.exe, eventvwr.exe, and sdclt.exe. Splunk also published supporting attack simulation data for suspicious child processes of consent.exe, giving defenders a way to test visibility for this privilege-escalation pattern, although the analytic is disabled by default and generates intermediate risk events rather than direct notable alerts.
Qakbot operators repeatedly changed their intrusion chains between March and May 2023, using malspam and a rotating set of attachment and container formats including PDF, HTML, ZIP, OneNote, WSF/HTA, and XLL files to deliver the malware. Researchers observed HTML smuggling, obfuscated JavaScript, PowerShell, and DLL-based payload staging in March, followed by ZIP-to-XLL chains that rebuilt payloads from split DAT files and created scheduled tasks for persistence. By April and May, the malware’s operators had shifted from base64-encoded PowerShell toward hex-encoded XMLHTTP requests delivered through WSF files, and used a OneNote-to-MSI lure posing as a Microsoft Azure installer. Later campaigns added stronger defense evasion, including indirect command execution through conhost.exe, DLL side-loading, and curl-based payload retrieval. Researchers also noted that Pikabot samples seen in mid-May shared similar tradecraft, but said the available evidence did not support attributing both malware families to the same actor.
Derp wouldn't exist without the work these projects do for the security community. We rely on their data sources to improve the quality and depth of what we publish. Thank you, we're genuinely grateful.