966 个 · 按相关度排序
Assess the security of LLM applications and AI agents by tracing inputs, retrieved content, tool calls, and outputs. Covers prompt injection, where content carries unwanted instructions, along with permission misuse, memory poisoning, and supply-chain risks.
中文简介正在准备中。当前显示英语简介。
Review GitHub Actions workflows for unsafe input handling, excessive token permissions, exposed secrets, and changeable action references. Get findings grouped by issue type, with severity ratings, exact YAML locations, and concrete fixes to review.
中文简介正在准备中。当前显示英语简介。
Monitor Modbus traffic in industrial control networks to identify unauthorized writes, unusual commands, and departures from normal communication. Use packet analysis and intrusion detection rules to investigate suspected controller changes and produce reports with recommended actions.
中文简介正在准备中。当前显示英语简介。
Add checks to LLM prompts and responses using Llama Guard 3, NeMo Guardrails, and LLM Guard. Configure content screening, conversation boundaries, and sensitive-data filtering, then measure blocking and false-positive rates with labeled test prompts.
中文简介正在准备中。当前显示英语简介。
Plan controlled failure experiments to uncover system weaknesses. Define normal operating metrics, estimate affected users and reliability risk, set abort and recovery conditions, and turn experiment results into a structured review with follow-up actions.
中文简介正在准备中。当前显示英语简介。
Assess security risks in AI systems by scanning prompts for known attack patterns and scoring data exposure, poisoning, and tool abuse risks. Map findings to MITRE ATLAS, an AI threat framework, and identify protective controls.
中文简介正在准备中。当前显示英语简介。
Review source code, dependencies, and infrastructure settings for security issues. Find exposed credentials, unsafe input handling, and access control gaps, then organize findings into a report with severity levels, affected locations, and recommended fixes.
中文简介正在准备中。当前显示英语简介。
Audits GitHub Actions workflows that run AI coding agents. Traces how attacker-controlled content reaches prompts, checks permissions and sandbox settings, and reports evidence, severity, and fixes for prompt injection and related configuration risks.
中文简介正在准备中。当前显示英语简介。
Plan persistence testing for authorized Windows and Linux security exercises, examining how access can survive logons or restarts. The skill compares mechanisms by privilege level and environment, with validation steps, detection indicators, and cleanup considerations.
中文简介正在准备中。当前显示英语简介。
Assess server-side template injection, where user input is interpreted as template code. Identify template engines, evaluate information exposure and code execution risks, and organize evidence alongside detection and prevention measures.
中文简介正在准备中。当前显示英语简介。
Assess weaknesses in the software supply chain, from dependency packages to build pipelines and distributed artifacts. Covers authorized proof-of-concept testing, package confusion, container image risks, signing abuse, and defensive checks for tampering.
中文简介正在准备中。当前显示英语简介。
Test LLM applications for malicious instructions, information leaks, and unsafe tool use. The skill runs attack cases and checks observable evidence, helping distinguish suspicious replies from reproducible failures with concrete impact.
中文简介正在准备中。当前显示英语简介。
Audit MCP servers, which connect AI agents to external tools, for injected instructions, excessive access, secret leaks, and unapproved actions. Trace untrusted input through tool calls and document demonstrated issues with concrete fixes.
中文简介正在准备中。当前显示英语简介。
Investigate cross-site scripting, where untrusted input executes as JavaScript in a web page. Check reflected, stored, and browser-side cases, distinguish escaped output from executable content, and validate reproducible behavior and user impact before reporting.
中文简介正在准备中。当前显示英语简介。
Investigate server-side template injection, where user input is interpreted as template code. The skill covers engine identification, testing email and PDF rendering, and checking whether an injection can lead to command execution on the server.
中文简介正在准备中。当前显示英语简介。
Assess Node.js applications for unsafe input handling and configuration weaknesses. The skill covers prototype pollution, where shared object properties are altered, plus template and command injection, proxy trust issues, and file exposure, with checks for observable impact.
中文简介正在准备中。当前显示英语简介。
Scan projects for known dependency vulnerabilities, exposed credentials, and risky code patterns such as SQL injection. Produces text or JSON reports with file locations and severity levels, with options to run individual checks.
中文简介正在准备中。当前显示英语简介。
Domain routing and boundary guidance for authorized SQL injection testing, including union-based, blind, error-based, stacked query, and second-order SQL injection variants. Use when a task belongs to the SQL injection domain and needs scope, evidence, pivot, or exit criteria.
中文简介正在准备中。当前显示原文说明。
Email header injection and spoofing playbook. Use when testing contact forms, email APIs, password reset flows, or any feature that constructs SMTP messages with user-controlled fields. Covers CRLF injection in headers, SPF/DKIM/DMARC bypass, and phishing amplification.
中文简介正在准备中。当前显示原文说明。
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.
中文简介正在准备中。当前显示原文说明。
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.
中文简介正在准备中。当前显示原文说明。
Detect GraphQL injection vulnerabilities in a codebase using a three-phase approach: recon (confirm GraphQL usage and find unsafe operation document assembly sites), batched verify (trace user input to those sites in parallel subagents, up to 3 candidate sites each), and merge (consolidate batch results). Requires sast/architecture.md (run sast-analysis first). Outputs findings to sast/graphql-results.md. If no GraphQL technology is found in Phase 1, later phases are skipped. Use when asked to find GraphQL injection, unsafe GraphQL document construction, or operation string injection bugs.
中文简介正在准备中。当前显示原文说明。
Audits legal AI skills, prompts, workflows, MCP/tool instructions, and agent packages for prompt injection, malicious instructions, unsafe scripts, suspicious metadata, credential exposure, exfiltration paths, persistence mechanisms, and supply-chain risk. Use before installing, importing, publishing, promoting, modifying, or trusting third-party or generated AI skills and legal workflows. Do not use as a generic code review unless skill trust, workflow safety, prompt injection, or supply-chain hygiene is in scope.
中文简介正在准备中。当前显示原文说明。
Use when managing dependencies between systems — autoloads, service locators, @export injection, and scene injection patterns
中文简介正在准备中。当前显示原文说明。