puntorigen/skills

14 skills47 total installs Github

Local media pipeline

Generate images, voiceovers, music, sound effects, radio drama, and talking-head video entirely on-device — no cloud APIs.

talking-head

bg-music

image-gen

sound-effects

voice-clone-narration

audio-theater

3D & web automation

Turn video into 3D Gaussian splats or printable 3D meshes, or teach the agent how a website works by recording sessions.

video-to-splat

object-to-3d

generate-web-skills

Documents

Read, edit, and create office documents locally — Word and PDF — while preserving formatting.

edit-docx

pdf-documents

Agent cognition

Give the agent a persistent, inspectable mental workspace — a local J-space concept graph it can load, query by association, prime, and checkpoint.

j-space

Brand & design

Generate a logo and a matching, brand-consistent asset set — local-first: it prefers on-device rendering via the image-gen skill (FLUX.2 Klein on Apple Silicon) and only falls back to a cloud Gemini/OpenRouter key (auto-discovered from your environment or another installed skill, never bundled) when local can't run. Brand text is set from real fonts via wordmark.py, and --look presets steer the visual finish.