Skill UI

A comprehensive client library for interacting with Azure Cosmos DB's NoSQL API in Rust. It enables performing core database operations, including CRUD (Create, Read, Update, Delete), patching, and managing globally distributed data. Ideal for building robust, data-intensive, and high-performance backend services on the Azure cloud platform.

Dispatch parallel agents to tackle independent failures concurrently so each investigator stays focused on one domain; review their summaries, check for conflicts, and run the full suite to integrate fixes faster than sequential debugging.

Guides multi-agent workflows through an iterative context retrieval loop—dispatching broad queries, evaluating relevance, refining criteria, and looping up to three times so sub-agents gather exactly the files they need without overwhelming limits.

Pyvene is a declarative framework designed for performing causal interventions on PyTorch neural networks. It allows researchers to conduct advanced experiments such as activation patching, causal tracing (ROME-style), and interchange intervention training (IIT). Use this library when you need to test causal hypotheses about model behavior, deeply interpret model components, or ensure reproducibility in advanced AI research.

A robust framework for orchestrating complex, multi-step coding tasks using multiple AI agents. It functions by first planning the entire project into a mission dependency graph (DAG), then dispatching specialized agents to isolated Git worktrees for parallel development. The system automatically monitors progress, handles merges, and generates structured, comprehensive reports detailing changes, completed tasks, and next steps.

A comprehensive guide detailing the systematic process of eradicating malware, backdoors, and attacker persistence mechanisms from compromised Windows and Linux systems. This workflow covers mapping persistence (Autoruns, registry, scheduled tasks), identifying artifacts (YARA, AV scans), removing threats, resetting credentials, and patching vulnerabilities to restore a clean state.

This comprehensive guide details the implementation of a Vulnerability Remediation SLA program. It covers establishing mandatory patching timeframes based on asset criticality and CVSS severity, defining exception processes, setting up escalation chains, and integrating tracking metrics (KPIs) into existing ticketing systems. Use this when defining mandatory security patching policies or maturing vulnerability management processes.

This skill provides a comprehensive framework for building an automated Service Level Agreement (SLA) breach alerting system for vulnerability remediation. It covers defining severity-based SLA policies (Critical, High, Medium, Low), structuring a database schema for tracking, implementing breach detection logic, and dispatching multi-channel escalation notifications (Email, Slack, PagerDuty). Essential for achieving compliance and improving overall security posture.

This harness transforms standard code environments into fully persistent, autonomous agent systems. It leverages native features like scheduled tasks (Crons), remote dispatching, and structured memory storage (MCP) to enable continuous, self-directing automation. Ideal for building AI assistants that require long-term context, recurring workflows, and complex, multi-step operations, eliminating the need for external agent frameworks like AutoGPT.

Dispatch dedicated agents with isolated contexts to handle multiple independent failures or tasks in parallel. Use when you have several unrelated issues that can be investigated separately without shared state, while preserving your own context for coordination and review.

This skill involves dispatching a code review sub-agent to validate work quality before merging or completing tasks. It helps identify issues early, ensuring critical and important problems are fixed. Use git SHAs to define the scope and follow feedback guidelines for critical, important, and minor issues.

This structured skill guides users through root-cause-first debugging when an implementer is blocked, verification fails, or repeated attempts at fixing do not converge. It mandates a systematic approach combining local evidence inspection, runtime analysis, and external documentation research to accurately classify the failure's true source (e.g., missing dependency, spec conflict). It prevents speculative patching and generates a clear, actionable fix plan.