avenoxai/avenoxskills

Produces brand-consistent motion graphics using HyperFrames for cutaways and overlays, then composites them onto video with ffmpeg. Used for title cards, data-viz, lower-thirds, and outros in video productions.

Automates rough cut of talking-head or screen recordings by removing silence and flubs/retakes using a transcript-driven pipeline; use when you need to cut dead air and bad takes.

A local-first video production pipeline for YouTube, using open-source tools (auto-editor, MLX-Whisper, MLT/melt, ffmpeg) and HyperFrames for motion graphics. Use when you need to edit, caption, score, assemble, or render a long-form video.

Query any major EVM chain's block explorer via the Etherscan V2 unified API. Look up contract ABI/source, transactions, receipts, logs, balances, token info, and verify contracts across 60+ chains.

Executes Codex CLI tasks, generates images with gpt-image-2, and orchestrates parallel fleets of codex delegates for high-throughput AI-driven code work and asset creation.

Delegation policy for agent stacks with a scarce main model and cheaper delegates. Routes context-gathering to Opus, implementation to Codex, and reserves judgment for the main loop. Use when spawning sub-agents or planning fleet work.

A disciplined workflow for sending coding tasks and audits to non-agentic frontier reasoning models (like GPT Pro) and verifying results against the live codebase before implementing. Use when you need deep reasoning on a static code snapshot.

Package a monorepo into clean, subsystem-split zip bundles for non-agentic LLMs like GPT Pro. Ideal for external code review, architecture audits, or feedback from models with bounded context.

Run parallel AI code tasks with omp CLI, leveraging your OpenAI Codex subscription. Supports cheap high-volume context gathering on a 25x cheaper model tier and fan-out orchestration.

Generates high-CTR YouTube thumbnails for mascot-driven channels, combining on-model mascot art with bold hooks and branded logos. Ideal for creating scroll-stopping covers and A/B testing thumbnails across video catalogs.