Skill UI

This skill implements Autonomous Optimization (AO) using Hypothesis Tree Refinement (HTR). It iteratively improves concrete artifacts (like code, models, or data pipelines) by running structured, long-term experiments against a measurable objective. It maintains a persistent hypothesis tree, coordinating subagents to test hypotheses in isolation while using a held-out test evaluator to prevent overfitting and ensure reliable, cumulative research progress. Ideal for complex benchmark tuning, agent refinement, or optimizing entire workflows.

This skill assists users in defining, adapting, and designing bounded, structured feedback loops for AI agents. Whether you need to find a suitable existing workflow, fine-tune a published loop, or design a new process from scratch, this tool guides you through a safe, iterative methodology. It emphasizes setting clear stopping rules, guardrails, and verifiable, measurable steps to ensure the automation remains controllable and focused.

A comprehensive security tool designed to detect and defend against subtle prompt injection attacks hidden within untrusted multi-modal content, such as web pages, PDFs, and images. It scans for obfuscated payloads (e.g., hidden CSS, metadata, zero-width characters) by combining content normalization, heuristic analysis, and dedicated ML detectors, ensuring that LLM agents only consume sanitized, trustworthy input.

This skill provides a comprehensive defense-in-depth framework for securing autonomous AI agents. It implements least-privilege tool allowlisting, identity binding, and human-in-the-loop controls to mitigate risks like prompt injection and tool poisoning. Use this when building agents that interact with sensitive systems (e.g., payments, file writes) to ensure governance, auditability, and compliance.