All skills

Finds what a change could break outside the diff, then proves the single fact it's safe because of by running real code.

Deep security auditing for git repos, AI skills and MCP servers — 28 scanners covering prompt injection, credential theft, supply-chain IOCs, and actively exploited CVEs.

Extracts characters, looks, locations, props and cross-scene continuity from a short-drama script into a reusable visual-setting document.

An agency-replacement skill that audits and implements five lanes — SEO, AEO, GEO, LLMO and Naver NEO — then sets up a content pipeline designed to get cited.

Paste a viral LinkedIn post URL and get back the hook formula it used, a structural breakdown, and a blank template for your own topic.

A hands-on guide to nanoGPT: train a tiny Shakespeare model on CPU or reproduce GPT-2 (124M) on multi-GPU, step by step.

A dev workflow skill that turns an approved feature spec into architecture, code, tests, and a QA handoff package.

Measures where GitHub Actions minutes and wait time actually go, then ships ranked fixes as small PRs with warm-cache before/after numbers.

Runs a new business, product, or side-project idea through a 9-dimension serial-founder filter and outputs a Build / Sleep on it / Pass viability brief.

Scores a live URL or local production build exactly as an Awwwards juror would — from screenshots and measured metrics only — and returns an evidence-backed score card with a ranked fix list.

Scans an Obsidian vault for unlinked mentions and inserts the missing [[wikilinks]] with confidence scoring and typed relationships.

Runs a daily Obsidian wiki maintenance pass (source freshness, index, hot.md, state files) and installs the 9 AM launchd/systemd/cron schedule.

Builds a ranked focus map of the load-bearing files, symbols and callers of an unfamiliar codebase, each with a file:line citation, before you change anything.

A multi-pass research workflow that plans sources, cross-checks them, and outputs a cited brief with contradictions and gaps — then archives every run for later grepping.

A mandatory QA gate that numerically validates Blender meshes (z-fighting, normals, non-manifold, UVs, weights, materials) and renders multi-angle review sheets before export or delivery.

A rulebook for building Blender 5.x extensions — manifest, registration, wheels, GPU drawing, CLI build and publishing.

A playbook for running multiple Claude Code sessions as a lead-plus-teammates team with shared tasks, messaging, and file locks.

Breaks a finished video into a per-shot analysis table (duration, shot size, category, camera move, framing) plus a single-file interactive shot-list report.

Mines Codex CLI session rollouts from ~/.codex and distills them into topic-based knowledge pages in an Obsidian vault.

Rebuilds a video with picture and sound only — no GPS, device IDs, chapters, telemetry or hidden encoder fingerprints — then proves it with a byte-level scan.