Results for “owasp-llm”

AI/LLM red-team assessment using the OWASP LLM Top 10 (2025) + OWASP AI Testing Guide (AITG v1, Nov 2025) frameworks, plus OWASP MCP Top 10 runtime testing for agentic/MCP targets. Tests prompt injection, jailbreaks, system prompt leakage, sensitive data extraction, excessive agency, improper output handling, model extraction, content bias, evasion, membership inference, MCP token exposure, MCP command injection, and more. Uses four tools in combination: FuzzyAI (single-turn jailbreak fuzzing), PyRIT (multi-turn orchestrated attacks), Garak (probe-based vulnerability scanning), and promptfoo (plugin-based red-team evaluation). Each tool covers different OWASP categories; running them together gives systematic coverage. Includes a conditional MCP reconnaissance phase and a post-access AI infrastructure phase (chained from /post-exploit). Produces: OWASP LLM Top 10 + AITG + MCP coverage matrix, findings per category, architecture diagram of the AI system, PoCs for confirmed exploits. Chains into /gh-export for

Read-only OWASP LLM Top 10 audit of app-facing AI features: prompt injection, data leak, supply chain, poisoning, unsafe output, excessive agency, system-prompt leak, RAG/embedding risks, misinformation, unbounded consumption. Use when "audit LLM security", "prompt injection", "jailbreak my chatbot", "is my AI safe".

Static OWASP review of app code (injection, headers, deps). Use when "review security" or "check vulnerabilities". Session/route×gate/getSession → audit-auth-flows. Plan-only burndown → plan-security-audit. Table RLS → plan-rls-audit. LLM attacks → audit-llm-security.

Security vulnerability scanner and OWASP compliance auditor for codebases. Dependency scanning (npm audit, pip-audit), secret detection (high-entropy strings, API keys), SAST for injection/XSS vulnerabilities, and security posture reports. Activate on 'security audit', 'vulnerability scan', 'OWASP', 'secret detection', 'dependency check', 'CVE', 'security review', 'penetration testing prep'. NOT for runtime WAF configuration (use infrastructure tools), network security/firewalls, or compliance certifications like SOC2/HIPAA (legal/organizational).

Deploy and serve LLMs with high throughput using vLLM's PagedAttention and continuous batching. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism for production inference.

Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.

Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.

Systematically audits a web application against the OWASP 2021 Top 10, producing severity-rated, file-level findings with fixes for each category.

Run Comet's Opik — open-source LLM observability, evaluation, and optimization — from one routing-first skill: install the Python/TypeScript SDK, stand up a server (Comet.com cloud, Docker Compose via `./opik.sh`, or Kubernetes/Helm), wire tracing through `@opik.track` or one of 50+ framework integrations (OpenAI, Anthropic, LangChain, LangGraph, LlamaIndex, CrewAI, DSPy, Ollama, Bedrock, Vercel AI SDK, …), score outputs with LLM-as-a-judge metrics (Hallucination, Moderation, Answer Relevance, Context Precision), and run Datasets/Experiments evaluations including PyTest CI gates. Use when the user wants LLM tracing, prompt evaluation, production LLM monitoring, agent optimization, or guardrails with Opik. Triggers on: opik, comet opik, opik configure, opik.sh, llm observability, llm tracing, llm as a judge, hallucination metric, prompt evaluation, opik dashboard, opik guardrails, agent optimizer.

Designing red team attack scenarios, threat models, MITRE ATT&CK/OWASP application, Purple Team exercises, and AI/LLM red teaming. Use when adversarial security validation is needed.

vLLM: high-throughput LLM serving, OpenAI API, quantization.

Scans a single file against the OWASP Top 10:2025, tracks cross-file connections, and maintains a persistent findings record.

Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.

Defensive review of an LLM/agent application for OWASP-LLM-style risks — prompt injection, unsafe tool calls, excessive agency, and PII/secret leakage. Reads your code, flags issues with file:line. Pure Claude Code, no signup.

Benchmark Jetson LLM/VLM serving performance across vLLM, llama.cpp, and Ollama with structured JSON output.

Program LM pipelines with DSPy — define Signatures (input/output fields), build Modules (ChainOfThought, ReAct, Predict), compose them into Programs, then optimize with Teleprompters (BootstrapFewShot, MIPRO, COPRO) to auto-improve prompts from examples.

Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.

Autonomous research review loop using any OpenAI-compatible LLM API. Configure via llm-chat MCP server or environment variables. Trigger with "auto review loop llm" or "llm review".

OWASP Top 10 audit checklists for Web Applications (2021), APIs (2023), and Mobile (2024). Use when performing any security review, PR review, or codebase audit touching web, mobile, or API code.

Enforces OWASP Top 10:2025 compliance on code touching security-relevant domains, using cached cheat sheets and verifying fixes against OSV.dev.

OWASP security guidelines and Top 10 vulnerabilities USE WHEN: user mentions "OWASP", "security audit", "vulnerability scan", asks about "injection", "XSS", "CSRF", "access control", "authentication security" DO NOT USE FOR: OWASP Top 10:2025 specific - use `owasp-top-10` instead

Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag). Use when benchmarking model quality, comparing models, reporting academic results, or tracking training progress. Industry standard used by EleutherAI, HuggingFace, and major labs. Supports HuggingFace, vLLM, APIs.

Integrate Uniswap liquidity provisioning (LP) into applications via the LP REST API. Use when the user says "LP API", "liquidity provisioning API", "provide liquidity programmatically", "create LP position via API", "add liquidity via API", "increase liquidity", "decrease liquidity", "remove liquidity", "claim LP fees", "collect LP fees", "manage LP positions in code", or mentions building a backend, bot, or frontend that creates or manages Uniswap v2/v3/v4 liquidity positions through an API. Also use when debugging LP API calls (e.g. /lp/create, /lp/check_approval, /lp/increase, /lp/decrease, /lp/claim_fees), unexpected response fields, the approval or EIP-712 permit flow, or transaction-building errors for liquidity positions. For generating deep links to the Uniswap web app instead of calling the API, use the liquidity-planner skill; for using the Uniswap v4 SDK directly rather than the REST API, use the v4-sdk-integration skill.

Track ML experiments with open-source run logging, local or self-hosted dashboards, and media visualization using SwanLab.

Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency on NVIDIA GPUs (A100/H100).

Language Server Protocol specialist building unified code intelligence systems through LSP client orchestration and semantic indexing

Composite skill — full security pass across secrets, dependencies, code paths, and OWASP risks. Chains security-audit (broad) + socket-audit (npm supply chain) + semgrep (pattern scan) + code-security (code review for vulns) in parallel, reconciles into one severity-ranked report with remediation plan. Use quarterly per active repo or before any release of security-sensitive code.

Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B.

Detect APT lateral movement by chaining Windows authentication events, process execution telemetry, and network connection logs across hosts using Splunk SPL and Sigma rule format.

Process textual and multimedia files with various LLM providers using the llm CLI. Supports both non-interactive and interactive modes with model selection, config persistence, and file input handling.