Skill UI

Vibers provides a human code review workflow for AI-generated GitHub projects; add the collaborator, GitHub Action, and commit guidance to receive spec- and security-focused feedback plus follow-up PR fixes.

This comprehensive guide provides a detailed workflow for parsing Windows LNK shortcut files in a digital forensics context. It extracts critical artifacts including target paths, timestamps (creation, access, modification), volume serial numbers, and machine identifiers. It is essential for reconstructing user file access history, tracking accessed files on network shares, and building robust forensic timelines.

Guides on using SwanLab for open-source ML experiment tracking, offering local or self-hosted dashboards, rich media logging, and integrations with PyTorch, Transformers, Lightning, or Fastai for comparing runs and monitoring metrics.

Guides fine-tuning and evaluation of OpenVLA-OFT/OFT+ for robot action generation, covering LoRA adaptation, continuous action heads, FiLM conditioning, and LIBERO/ALOHA benchmarks plus deployment considerations.

Provides end-to-end guidance for fine-tuning and serving Physical Intelligence’s OpenPI pi0 family (pi0, pi0-fast, pi0.5) with both JAX and PyTorch backends, covering dataset prep, norm stats, checkpoint conversion, and low-latency WebSocket policy inference for ALOHA, DROID, and LIBERO tasks.

Hard screening skill runs deterministic VC analyses where Claude gathers context, Python scores/formats, and the pipeline produces auditable verdicts with investment thesis guidance.

This guide details implementing an IEC 62443 compliant conduit architecture for secure remote access into Operational Technology (OT) environments. It mandates the use of jump servers, MFA gateways, session recording, and approval workflows within a DMZ to strictly control vendor and engineer access, eliminating direct VPN connections to critical ICS assets.

Guides users through a structured 23-step interview to design, generate, and deploy Claude plugins across claude.ai and Claude Code, handling activation prompts, platform quirks, and GitHub pushes.

Soft Predict Future is a structured foresight engine that triggers whenever the user asks about future outcomes or scenarios. It enforces a nine-step analytical pipeline (validation, signal collection, scoring, driver extraction, etc.), highlights required searches and inputs, and directs the user to web-enabled reasoning for predictive analysis.

Guides analysts through soft-mode startup screening by extracting company profiles, scoring eight VC dimensions, applying four investor lenses, and delivering a PASS/DECLINE verdict with a concise memo.

The /hub:spawn command launches N subagents that work on the same task simultaneously. Each agent operates within an isolated git worktree, ensuring true parallelism and resource separation. Users can optionally specify templates (e.g., `optimizer`, `refactorer`) to guide agents toward specific strategies for improving code quality, performance, or test coverage. Ideal for complex software engineering tasks requiring diverse, competing approaches.

Automatically analyzes unfamiliar codebases to generate a comprehensive, structured onboarding guide. It systematically detects package manifests, framework fingerprints, architecture maps, key entry points, and coding conventions. Ideal for new developers joining a project or setting up a repository for the first time.