AI skill directory

Operates the Hugging Face Hub CLI for downloads, uploads, repositories, papers, jobs, and Spaces.

Fine-tunes language and vision models with TRL or Unsloth on Hugging Face Jobs.

Routes video, animation, and motion-graphics requests to the appropriate HyperFrames workflow.

Runs the HyperFrames CLI workflow for initialization, preview, rendering, publishing, linting, and diagnostics.

Designs, critiques, and polishes interfaces through audit, clarification, distillation, and craft reviews.

Implements a scoped piece of work from a spec or tickets with a disciplined execution loop.

Reviews and improves codebase architecture—boundaries, modules, and coupling—before large refactors.

Designs raw mechanical interfaces using Swiss typographic grids and military-terminal aesthetics.

Helps write company-style internal communications—status updates, newsletters, FAQs, and similar formats.

Writes customer interview scripts tailored to discovery goals and bias-aware questioning.

Investigates bugs and product issues with a structured evidence-gathering loop before proposing fixes.

Applies Jobs-to-be-Done framing to uncover customer jobs, forces of progress, and solution fit.

Creates and edits Jupyter notebooks for analysis, demos, and reproducible research workflows.

Plans product launches—messaging, channels, checklist, and launch-day coordination.

Builds high-risk features as lean slices grounded in the repository's existing structure.

Fills a Lean UX Canvas to align business outcomes, users, hypotheses, and experiment design.

Creates and manages Linear issues, projects, and workflow updates from agent sessions.

This skill makes animated GIFs that fit Slack, at emoji or message size, with frame rate, color, and length limits. It draws frames from shapes or a provided image, saves an optimized file, and checks it against Slack's limits.

This skill creates YouTube thumbnails and vertical covers for Shorts or Instagram through the Higgsfield CLI. It starts from an honest concept for the video, can keep up to three referenced faces, and can add a logo or text on top. Small fixes after the main render are done as targeted edits. It does not make the video.

This skill works with Gmail through the Google Workspace CLI (gws). Helper commands send, read, reply, forward, and watch for new mail, and one lists unread messages for triage. Raw Gmail API calls are available for anything else.