Skill UI

This module provides comprehensive abstractions, data types, and utilities for managing webhooks. It includes interfaces for defining payloads, tracking delivery status, dispatching events, and verifying incoming webhooks, enabling robust event-driven integration between services.

A comprehensive security skill designed to detect and mitigate indirect prompt injection attacks. It scans untrusted external content—including web pages, PDFs, and images—for hidden, obfuscated instructions before an LLM agent processes them. The pipeline combines content extraction (HTML, OCR), normalization, and dedicated ML models (e.g., LLM Guard, Prompt Guard 2) to ensure data safety.

RouterBase acts as a unified, OpenAI-compatible API gateway, allowing applications to route requests across diverse large language models (LLMs) like GPT, Claude, and Gemini. It helps developers migrate existing OpenAI SDK calls, implement sophisticated primary/fallback model selection based on cost, latency, and quality, and manage multimodal payloads (image, audio, video) from a single, standardized endpoint.

This tool performs comprehensive safety scanning on untrusted inputs, such as user submissions, API payloads, or webhook data. It detects sophisticated threats like prompt injection, jailbreak attempts, and adversarial attacks using advanced AI defense mechanisms, ensuring data integrity and security before model processing or execution.

A specialized downstream skill for complex CTF challenges, focusing on cryptographic analysis, steganography, and mobile system forensics. It guides the user through recovering complex data chains, such as decoding blobs, tracing cryptographic transforms (keys, IVs, nonces), inspecting hidden media payloads, and analyzing mobile trust boundaries (e.g., APK/IPA signing logic, local storage, SSL pinning). This skill is designed to be used only after a main sandbox orchestrator has established the operational context.

A highly specialized, downstream skill for advanced threat analysis. This tool is used to recover hidden configurations, decode C2 (Command and Control) parameters, unpack staged payloads, and extract Indicators of Compromise (IOCs) from complex malware samples. It focuses on identifying the exact decode chains—including compression, encoding, and crypto steps—and linking recovered fields to observed behavioral outcomes within a sandbox environment.

This skill is designed for deep forensic analysis of asynchronous processing chains, specifically focusing on 'drift'—behavioral or configuration differences between a synchronous request path and subsequent worker-executed tasks. Use it to trace payloads, inspect worker-only state, follow retry mechanisms, handle dead-letter queues, or connect enqueued items to eventual side effects (e.g., file writes, cache updates, emails). It is a specialized downstream tool, requiring prior orchestration context.

This advanced skill specializes in detecting and recovering hidden data (steganography) embedded within various media containers, including images, audio, video, and documents. It is crucial for forensic analysis and CTF challenges, allowing users to inspect metadata, LSBs, alpha/palette channels, and appended trailers to reconstruct hidden payloads without brute force.

This guide details Phase 4: transitioning an interactive agent into a persistent, automated workflow. It covers creating scheduled deployments (cron jobs) or event-driven triggers (cURL). Users learn to validate cron schedules, build the necessary deployment payloads, and follow best practices, including manual testing and DST consideration, ensuring the agent runs reliably without human intervention.

This skill automates the final deployment phase of an agent system. It converts a validated build sheet into four ordered JSON payloads (environment, agent, session, kickoff) and generates a resumable `launch.sh` script. Crucially, this process enforces API-key safety by requiring the user to run the script in their shell, pulling the Anthropic API key from the environment variable, thus ensuring the key never appears in logs or files. It provides a structured, multi-step launch sequence.

A comprehensive, black-box workflow designed for professional bug bounty hunting and penetration testing. It guides users through a 5-stage methodology (Intake, Recon, Enum, Hunt, Report), covering 19+ advanced attack playbooks (e.g., RCE, SQLi, XSS, SSRF). The skill integrates real-world exploit cases, structured payloads, and WAF bypass techniques for maximizing vulnerability discovery in web applications and APIs.

Performs non-intrusive static analysis of AI Skills and Bundles to detect malicious patterns, data leaks, privilege escalation, obfuscated payloads, persistence and mobile threats across Windows, macOS, Linux/Unix, Android and iOS, producing evidence-based audit reports.