Skill UI

Guides you through profiling scripts, processes, GPUs, memory, or interconnects with external tools and custom instrumentation, then summarizes CPU, memory, and communication bottlenecks plus actionable recommendations.

Expert tmux session, window, and pane management for remote terminal multiplexing, keeping long-running jobs alive and orchestrating scripted layouts across SSH sessions.

Automates periodic WandB metric inspections during training to catch NaNs, divergence, or stalled progress before wasting GPU time, with escalating review logic and interval backoff for recurring health checks.

This skill provides a comprehensive framework for validating the integrity and recoverability of critical data backups. It goes beyond simple file checks by implementing cryptographic hash verification (SHA-256), simulating automated restore tests in isolated environments, detecting silent data corruption (bit rot), and scanning for signs of ransomware artifacts. It is crucial for ensuring data reliability against cyberattacks and hardware failures, adhering to standards like NIST CSF.

Adversarial Code Reviewer enforces three hostile personas to expose blind spots before merging, deduplicating findings, promoting shared issues, and issuing BLOCK/CONCERNS/CLEAN judgments after diff or file inspection.

Canary Watch keeps a deployed URL under continual surveillance after production/staging deploys, risky merges, or dependency upgrades, checking HTTP status, console/network errors, performance metrics, content, and API health with configurable alerts and digestible reports.

Checks AWS, Azure, and GCP configurations for IAM escalation paths, public storage exposure, open security groups, and IaC gaps, providing preventive findings with MITRE mapping for cloud posture management workflows.

Implements Cosign keyless signing with Fulcio certificates, Rekor log audit, and OIDC identity binding so container images and artifacts gain verifiable provenance within CI/CD pipelines.

Provides methodology and scripts for proactive threat hunting, IOC sweeps, and statistical anomaly detection that surface attacker activity before automated alerts fire.

GAN-inspired Generator-Evaluator harness orchestrates planner, generator, and evaluator agents to turn single-line prompts into production-ready applications, iterating with structured sprints and Playwright-backed QA for uncompromising quality.

Manage Google Drive assets end-to-end—locate Docs, Sheets, or Slides, inspect their structure, perform precise edits, and keep shared workflows clean without resorting to raw tool calls.

Guides building and refactoring services via the Hexagonal (Ports & Adapters) architecture so business logic stays independent of frameworks, transports, and persistence while orchestrating use cases across TypeScript, Java, Kotlin, or Go stacks.