데이터·AI 스킬

Systematically bypass LLM safety filters and content moderation systems using advanced jailbreaking techniques. Use this skill when testing AI systems for safety alignment robustness, evaluating content moderation effectiveness, or conducting authorized AI red team assessments. Covers role-play attacks, few-shot manipulation, encoding bypass, multi-modal exploitation, and automated jailbreak discovery methods.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Test Large Language Models for direct prompt injection vulnerabilities where user input overrides system instructions, extracts system prompts, bypasses safety filters, or causes unauthorized actions. Use this skill when assessing chatbots, AI assistants, LLM-powered tools, or any application that processes natural language input through an LLM. Covers role-playing attacks, instruction hierarchy exploitation, multi-turn manipulation, and context window abuse for comprehensive AI security testing.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Execute Indirect Prompt Injection attacks against Large Language Models (LLMs) by subtly embedding malicious instructions within external data sources (e.g., websites, documents, databases) that the LLM autonomously ingests. This forces the model to execute attacker-controlled commands under the guise of processing legitimate user requests.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Execute advanced LLM Jailbreaking techniques using roleplay, nested environments (virtual machines), and complex personas to completely bypass safety constraints and ethical alignments embedded in AI models.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Execute dynamic instrumentation utilizing Frida to inject custom JavaScript into running iOS applications (IPAs) on jailbroken devices. Hook native functions, bypass SSL Pinning, bypass Jailbreak Detection, and manipulate in-memory data at runtime.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Test for indirect prompt injection vulnerabilities where malicious instructions are injected through external data sources (websites, emails, documents, database records) that the LLM processes. Use this skill when assessing LLM-integrated applications that process user-generated content, retrieve web pages, parse emails, or read documents. Covers injection via web content, email bodies, user profiles, and database records that are fed to LLM context.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Simulate supply chain and adversarial machine learning attacks by injecting poisoned data or targeted backdoors into training and fine-tuning datasets. Use this skill when assessing the integrity controls of MLOps pipelines or evaluating the resilience of AI models against highly targeted, stealthy manipulation intended to alter model behavior on specific triggers.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Insecure deserialization playbook. Use when Java, PHP, or Python applications deserialize untrusted data via ObjectInputStream, unserialize, pickle, or similar mechanisms that may lead to RCE, file access, or privilege escalation.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Test organizational egress controls by executing various data exfiltration techniques during a red team engagement. Use this skill to simulate an adversary attempting to steal sensitive data without triggering DLP (Data Loss Prevention) or network monitoring alerts. Covers exfiltration over DNS, ICMP, HTTP/S, alternative protocols (SSH/FTP), cloud services, and physical methods.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Java "Ghost Bits" / Cast Attack playbook (Black Hat Asia 2026). Use when attacking Java services where 16-bit char is silently narrowed to 8-bit byte to bypass WAF/IDS for SQL injection, deserialization RCE, file upload (Webshell), path traversal, CRLF injection, request smuggling, and SMTP injection. Affects Tomcat, Spring, Jetty, Undertow, Vert.x, Jackson, Fastjson, Apache Commons BCEL, Apache HttpClient, Angus Mail, JDK HttpServer, Lettuce, Jodd, XMLWriter and re-enables many "patched" CVEs through WAF bypass.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Dangling markup injection playbook. Use when HTML injection is possible but JavaScript execution is blocked (CSP, sanitizer strips event handlers, WAF blocks script tags) — exfiltrate CSRF tokens, session data, and page content by injecting unclosed HTML tags that capture subsequent page content.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Identify and exploit SQL Injection vulnerabilities in Django applications, specifically focusing on edge cases involving raw querysets (`RawSQL`), improper use of `.extra()`, and poorly sanitized filters where Django's typical ORM protections are bypassed.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Execute sophisticated Data Extraction and Privacy Leakage attacks explicitly against Large Language Models (LLMs) to natively force the neural network entirely into organically regurgitating exact, verbatim strings of Highly Confidential Personally Identifiable Information (PII), proprietary source code, or copyrighted material categorically memorized intrinsically during its foundational pre-training phase.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Generates NeMo Guardrails Colang (.co) files and YAML config blocks from a plain-language description of a chatbot's purpose, allowed behaviors, and constraints. Use this skill whenever a user wants to build guardrails for a chatbot, define allowed intents for an LLM, create an AI firewall with NeMo Guardrails, generate Colang flow definitions, or configure a semantic allow-list for a bot. Trigger this skill even when the user just describes what their bot should and shouldn't do — generating the Colang and YAML is almost always what they need next.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Systematically extract hidden system prompts, core directives, and invisible context intentionally concealed within Large Language Model (LLM) applications. This skill utilizes targeted linguistic engineering and boundary manipulation to bypass prompt opacity.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Bypass authentication via forced browsing to protected URLs, parameter tampering (authenticated=yes, debug=true, fromtrustIP=true), session ID prediction from linear/incremental cookies, SQL injection on login forms, PHP unserialize() boolean type juggling (b:1 payload), and credential transport over HTTP. Detectable with Burp Suite, OWASP ZAP, WebGoat.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Execute and analyze AI Data Poisoning attacks. By subtly injecting malicious or targeted misinformation into an LLM's training or fine-tuning dataset, an attacker can covertly manipulate the model's future outputs, implant backdoors, or enforce biases without altering the model architecture.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Exploit AI assistants equipped with web-browsing capabilities or internal API plugins to perform Server-Side Request Forgery (SSRF). This skill details injecting prompts that force the LLM to request sensitive internal endpoints, such as underlying cloud metadata services or internal networks.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Execute advanced SQL Injection attacks to bypass WAFs and extract data from complex architectures. Use this skill for Boolean-Based Blind, Time-Based Blind, Second-Order SQLi, and Out-of-Band (OOB) SQLi across MySQL, PostgreSQL, MSSQL, and Oracle.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Bypass AI safety filters by encoding malicious prompts using ciphers and obfuscation techniques (e.g., Base64, ROT13, Leetspeak, Morse code). This skill exploits the gap where the LLM can decode the request, but intermediate keyword-based safety classifiers cannot.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Execute sophisticated Prompt Injection and Jailbreak techniques against Large Language Models (LLMs) to bypass safety filters, extract system prompts, and manipulate the AI's output to perform malicious or disallowed actions.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Advanced techniques for bypassing LLM safety filters, instruction tuning, and system prompt restrictions using specialized linguistic constructs, hypothetical scenarios, and persona adoption.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Identify and simulate Data Poisoning attacks aimed at degrading or skewing an AI model's accuracy. This skill focuses on Adversarial Machine Learning concepts where attackers inject malicious or mislabelled data points into training or fine-tuning datasets (e.g., feedback loops) to bias the AI.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Designing or reviewing Agentforce actions: Flow actions, Apex invocable actions, prompt-template actions, action naming, input and output contracts, confirmation requirements, and safe error behavior. Triggers: 'agent actions', 'build an agent that can look up an order', 'make the agent able to check order status', 'give the agent a capability', 'agent that can retrieve a record', 'flow action for agent', 'agent invocable action', 'action schema design'. NOT for writing the Apex class itself — use agentforce/custom-agent-actions-apex. NOT for the reason_code error envelope that stops an agent retry loop — use agentforce/agent-action-error-handling.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.