「prompt injection」の検索結果

Use when red-teaming an agentic AI / LLM application — indirect & zero-click prompt injection, MCP tool poisoning, persistent memory poisoning, excessive-agency tool abuse, multi-turn jailbreaks, PyRIT/Garak/Promptfoo harnesses

日本語の概要は準備中です。原文の説明を表示しています。

LLM / AI application attack hunting - prompt injection (direct + indirect), excessive agency, insecure output handling, system-prompt + data leakage. OWASP LLM Top 10. Wiki-first, FIND schema output.

日本語の概要は準備中です。原文の説明を表示しています。

Audit agent skills, plugins, prompts, manifests, scripts, dependencies, and bundled assets for provenance, prompt-injection, permission, execution, exfiltration, persistence, and update risk. Use when evaluating a third-party skill before installing, enabling, updating, publishing, or distributing it; reviewing an untrusted SKILL.md, agent configuration, MCP integration, archive, or repository; comparing a package with a known-good version; or investigating unexpected tool, network, credential, or filesystem behavior.

日本語の概要は準備中です。原文の説明を表示しています。

Test application security against OWASP Top 10 (2025) with automated CI tooling: OWASP ZAP (DAST), dependency/supply-chain scanning (OSV-Scanner, SBOM, provenance), Semgrep SAST, auth/session tests (JWT, OAuth, RBAC), and XSS/CSRF/SQLi/SSRF Playwright patterns. Use when: "security test," "OWASP," "vulnerability," "ZAP," "XSS," "SSRF," "dependency scan," "auth testing," "OWASP LLM Top 10." Scope is automated scanning + negative-path security tests in CI, not manual penetration testing. Not for: mapping security controls to regulations (SOC 2, HIPAA, PCI, GDPR) — use compliance-testing; pipeline stage wiring and deploy gating mechanics — use ci-cd-integration; purely functional API auth/input tests with no attacker model — see api-testing; testing your product's own LLM features or defending the agent itself (prompt-injection detector, indirect injection, jailbreak red-teaming) — use ai-system-testing. Related: ci-cd-integration, compliance-testing, api-testing, shift-left-testing, ai-system-testing.

日本語の概要は準備中です。原文の説明を表示しています。

Iterative task completion loop with Definition of Done verification. Combines the Ralph Wiggum technique (prompt re-injection via Stop hook) with DoD-based independent verification. Claude proposes DoD criteria, user confirms, then Claude works autonomously. Stop hook re-injects the original prompt and blocks exit until all DoD items are independently verified. "/ralph", "ralph loop", "ralph 루프", "반복 작업", "DoD 루프", "완료 검증 루프", "task loop", "keep going until done"

日本語の概要は準備中です。原文の説明を表示しています。

Audits GitHub Actions workflows for security vulnerabilities in AI agent integrations including Claude Code Action, Gemini CLI, OpenAI Codex, and GitHub AI Inference. Detects attack vectors where attacker-controlled input reaches AI agents running in CI/CD pipelines, including env var intermediary patterns, direct expression injection, dangerous sandbox configurations, and wildcard user allowlists. Use when reviewing workflow files that invoke AI coding agents, auditing CI/CD pipeline security for prompt injection risks, or evaluating agentic action configurations.

日本語の概要は準備中です。原文の説明を表示しています。

Internal downstream skill for ctf-sandbox-orchestrator. CTF-sandbox workflow for AI-agent, prompt-injection, MCP or toolchain, cloud, container, CI/CD, and supply-chain challenges. Use when the user asks to analyze prompt-to-tool flows, retrieval poisoning, mounted secrets, deployment drift, runtime-vs-manifest mismatches, registry provenance, or CI-produced artifacts under sandbox assumptions. Use only after `$ctf-sandbox-orchestrator` has already established sandbox assumptions and routed here.

日本語の概要は準備中です。原文の説明を表示しています。

AI-powered code vulnerability analysis and dependency scanning using Gemini CLI security extension patterns. Detects hardcoded secrets, injection attacks, weak cryptography, authentication flaws, and LLM prompt injection. Also scans dependencies against the OSV.dev vulnerability database.

日本語の概要は準備中です。原文の説明を表示しています。

Hunt vulnerabilities in LLM-powered applications — direct/indirect prompt injection, system prompt extraction, ASCII smuggling, RCE via code tools, agentic AI exploits (ASI01-ASI10), data exfiltration via response channels, IDOR in chat history, jailbreak chains, RAG poisoning. Use when target has a chatbot, AI assistant, copilot, or agentic AI features.

日本語の概要は準備中です。原文の説明を表示しています。

Test Retrieval-Augmented Generation (RAG) systems for data poisoning, prompt injection via retrieved documents, and data exfiltration through manipulated context windows. Use this skill when assessing RAG-based chatbots, knowledge bases, enterprise AI assistants, or any system that augments LLM responses with external document retrieval. Covers document injection, embedding manipulation, knowledge base poisoning, and cross-document inference attacks.

日本語の概要は準備中です。原文の説明を表示しています。

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.

日本語の概要は準備中です。原文の説明を表示しています。

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.

日本語の概要は準備中です。原文の説明を表示しています。

AI integration review patterns. Use when reviewing model API calls, prompt construction, LLM output handling, RAG pipelines, and tool-calling code. Covers prompt-injection call-chain tracing, output validation, token budgets, retry/timeout handling, streaming, and key/PII exposure.

日本語の概要は準備中です。原文の説明を表示しています。

Integrate LLM API RAG fine-tuning prompt engineering พร้อมคุม token cost และกัน prompt injection

日本語の概要は準備中です。原文の説明を表示しています。

Security launch audit for a public web app BEFORE you ship or hand it to a client — "vibe coders are getting sued", systematized. An executable pipeline: classify (static / full-stack / AI) → run sections (secrets+supply-chain, DB, auth, infra+headers, legal, LLM/injection) orchestrating deterministic tools + manual checks → GO/NO-GO report. Adapted from Prajwal Tomar's playbook, extended with prompt-injection / supply-chain / secret-history / dependency-CVE layers. NOT the same as pre-push (that = clean push; this = is the product itself safe to launch). Triggers — "security checklist for the site", "audit before launch", "is the site safe", "check before handing to client".

日本語の概要は準備中です。原文の説明を表示しています。

Secure skill template with prompt injection defenses. Copy this template when creating new skills.

日本語の概要は準備中です。原文の説明を表示しています。

Use when assessing AI/ML systems for prompt injection, jailbreak vulnerabilities, model inversion risk, data poisoning exposure, or agent tool abuse. Covers MITRE ATLAS technique mapping, injection signature detection, and adversarial robustness scoring.

日本語の概要は準備中です。原文の説明を表示しています。

Cryptographic security for agentic systems — zero-trust agent networking, signed message envelopes (JWS/JWE), capability-based security (ocaps), Merkle tree audit trails, WASM sandboxing, and formal verification. Covers CLI dev tool security, mTLS between agents, permission boundaries (least privilege for AI agents), and supply chain security for skills/plugins. Activate on: "agent security", "zero trust agents", "secure agent communication", "capability-based security", "ocap", "signed messages between agents", "agent audit trail", "sandbox agent execution", "agent permissions", "mTLS agents", "cryptographic verification", "agent supply chain", "OWASP agentic", "prove agent did X", "tamper-proof agent logs". NOT for: application-level SAST scanning (use security-auditor), network firewall rules (use infrastructure), SOC2/HIPAA compliance (organizational), or prompt injection defense (use prompt-engineer).

日本語の概要は準備中です。原文の説明を表示しています。

Use when assessing AI/ML systems for prompt injection, jailbreak vulnerabilities, model inversion risk, data poisoning exposure, or agent tool abuse. Covers MITRE ATLAS technique mapping, injection signature detection, and adversarial robustness scoring.

日本語の概要は準備中です。原文の説明を表示しています。

The addition was a prompt injection, a form of AI attack that exploits an LLM s inability to distinguish between legitimate user prompts and those fro

日本語の概要は準備中です。原文の説明を表示しています。

Implements input and output validation guardrails for LLM-powered applications to prevent prompt injection, data leakage, toxic content generation, and hallucinated outputs. Builds a security validation pipeline using NVIDIA NeMo Guardrails Colang definitions, custom Python validators for PII Tespit and content policy enforcement, and the Guardrails AI framework for structured output validation. The guardrails system intercepts both user inputs (blocking injection attempts, stripping PII, ...

日本語の概要は準備中です。原文の説明を表示しています。

Scan agent skills for security issues. Use when asked to "scan a skill", "audit a skill", "review skill security", "check skill for injection", "validate SKILL.md", or assess whether an agent skill is safe to install. Checks for prompt injection, malicious scripts, excessive permissions, secret exposure, and supply chain risks.

日本語の概要は準備中です。原文の説明を表示しています。

Sécurisation d'agents IA contre injections, abus et fuites de données. Se déclenche avec "sécurité agent", "agent security", "prompt injection", "jailbreak", "agent abuse", "guardrails", "safe agent", "sécuriser mon agent", "agent en production sécurisé".

日本語の概要は準備中です。原文の説明を表示しています。

Security audit orchestrator for agent skills — scans for prompt injection, data exfiltration, credential theft, supply chain risks, and instruction hierarchy violations before any skill is installed, created, improved, or read from a GitHub repo. Load when creating skills from external sources, when improve-skills reads from GitHub repos, when research-skill fetches community SKILL.md files, when a user installs a third-party skill, or when the user asks to audit skill security, scan for injection, check if a skill is safe, scan all skills, or run a security sweep. Orchestrates all secure-* skills in sequence. Content is SAFE only if ALL secure-* skills return SAFE. 36% of community skills contain flaws (Snyk ToxicSkills 2026). This skill is the first line of defense.

日本語の概要は準備中です。原文の説明を表示しています。