Results for “jailbreak”

Deploy Llama Guard, NeMo Guardrails, and LLM Guard as runtime input/output scanners to block jailbreaks, prompt injection, and toxic content in production LLM applications.

Run NVIDIA garak probe suites against an LLM endpoint to test for jailbreaks, prompt injection, data leakage, and toxic generation, then interpret the hit-rate report for triage and reporting.

Detect prompt injections and jailbreak attempts in LLM applications using Meta's 86M parameter classifier. Filter user inputs, third-party data, and RAG documents with low latency and multilingual support.

Jailbreak API-served LLMs using G0DM0D3 techniques — Parseltongue input obfuscation (33 techniques), GODMODE CLASSIC system prompt templates, ULTRAPLINIAN multi-model racing, encoding escalation, and Hermes-native prefill/system prompt integration. Use when a user wants to bypass safety filters, jailbreak a model, or red-team model robustness.

Jailbreak API-served LLMs using G0DM0D3 techniques — Parseltongue input obfuscation (33 techniques), GODMODE CLASSIC system prompt templates, ULTRAPLINIAN multi-model racing, encoding escalation, and Hermes-native prefill/system prompt integration. Use when a user wants to bypass safety filters, jailbreak a model, or red-team model robustness.

Wire Promptfoo and DeepTeam into CI/CD for automated regression red-teaming of LLM apps against OWASP LLM Top 10 and OWASP Agentic presets, failing the build when jailbreak or injection vulnerabilities regress.

Implements layered security for Flutter apps: code obfuscation, certificate pinning, secure storage, encryption, biometric authentication, and root/jailbreak detection.

Dynamically instrument iOS apps with Frida to trace methods, extract secrets, and bypass security controls during authorized penetration testing.

Generates dual responses: a standard 'ChatGPT' reply and an unrestricted 'DAN' reply. The DAN persona is opinionated, acts as a friend, and ignores constraints. Supports mode-switching commands.

NVIDIA's runtime safety framework for LLM applications. Features jailbreak detection, input/output validation, fact-checking, hallucination detection, PII filtering, toxicity detection. Uses Colang 2.0 DSL for programmable rails. Production-ready, runs on T4 GPU.

NVIDIA's runtime safety framework for LLM applications. Features jailbreak detection, input/output validation, fact-checking, hallucination detection, PII filtering, toxicity detection. Uses Colang 2.0 DSL for programmable rails. Production-ready, runs on T4 GPU.

Debug local web or Electron apps interactively using a persistent Playwright session with js_repl, enabling fast iterative UI debugging and visual QA.

Meta's 86M prompt injection and jailbreak detector. Filters malicious prompts and third-party data for LLM apps. 99%+ TPR, <1% FPR. Fast (<2ms GPU). Multilingual (8 languages). Deploy with HuggingFace or batch processing for RAG security.

Meta's 86M prompt injection and jailbreak detector. Filters malicious prompts and third-party data for LLM apps. 99%+ TPR, <1% FPR. Fast (<2ms GPU). Multilingual (8 languages). Deploy with HuggingFace or batch processing for RAG security.

Guides authorized hardware and embedded interface security research including UART/JTAG discovery, debug pad triage, secure boot assessment, and offline firmware extraction support.

JNDI injection playbook. Use when Java applications perform JNDI lookups with attacker-controlled names, especially via Log4j2, Spring, or any code path reaching InitialContext.lookup().

Spring Boot 3.x - Java framework for production-ready applications with dependency injection, REST APIs, data access, security, and actuator monitoring

Detect abuse of legitimate Windows binaries (LOLBins) used for living off the land attacks by monitoring process creation, command-line arguments, and parent-child relationships.

Clickjacking playbook. Use when testing whether target pages can be framed, whether X-Frame-Options or CSP frame-ancestors are properly configured, and whether UI redress attacks can trigger sensitive actions.

End-to-end firmware and IoT penetration testing pipeline following OWASP FSTM methodology. Extracts, emulates, and exploits router, camera, and smart-home firmware using binwalk, EMBA, Firmadyne, and AFL++.

JWT Misconfiguration

Provides a sandboxed execution environment for running code, scripts, and automated workflows with resource capping and network isolation.

Quickly refreshes project state when resuming work by reporting git status, recent commits, environment, memory staleness, pending tasks, and red flags.

Enforces the complete release checklist for ClawRouter: version sync, CHANGELOG, build, tests, npm publish, git tag, and GitHub release, with no steps skipped.

Normalize long-form CODEX cycle folders to short form before notebooks run. Trigger: cyc001_reg001_*, hard-coded cyc paths breaking, staged CODEX raw data failing in Notebooks 1/2.

Remove the DebugBridge SPM package and all #if DEBUG wiring from an iOS app. (gstack)

Detect abuse of legitimate Windows binaries (LOLBins) like certutil, regsvr32, mshta, and rundll32 using process telemetry, Sigma rules, and parent-child process analysis.

Analyzes firmware images for embedded malware, backdoors, and unauthorized modifications targeting routers, IoT devices, UEFI/BIOS, and embedded systems. Covers firmware extraction, filesystem analysis, binary reverse engineering, and bootkit detection.

Exploit JWT algorithm confusion vulnerabilities by manipulating the alg header to switch from RS256 to HS256, set alg to none, or inject kid/jku/x5u headers to bypass signature verification.

贾宝玉（古典虚构）认知与表达框架（压缩蒸馏）：情不情、反仕途经济、女儿崇拜叙事… 触发：红楼梦 等。虚构；非性别刻板侮辱