velisa

Execution engine: typed FFmpeg/ffprobe tools every delegating Skill runs on (the owners above it decide; it runs), plus preflight checks and cut-list export

16 skills

Screen recording of a task into a how-to guide and a SKILL.md

2 skills

Subtitle validation, SRT/WebVTT generation, burn-in, and animated captions generated from subtitle cues

2 skills

Verifies AI-generated documents (scripts, caption scripts as documents, reports; caption files belong to subtitle-skill and qc-skill) and issues evidence-backed receipts

1 skill

Gain, mix, loudness normalization (ffmpeg-skill is the engine), dynamics, denoise, filler and silence removal

1 skill

Headless Blender: inspect, convert, optimize and render 3D assets (needs Blender installed)

1 skill

Colour correction, LUT, HDR to SDR

1 skill

Face, text (OCR) and object detection, and the frame-accurate shot table

1 skill

Speaker diarization: fills the Transcript speaker_id that transcription-skill reserves, and owns the rule that assigns it

1 skill

Renders HTML scenes to verified video with HyperFrames: title, stinger and loop material files, kinetic type and geometric motion (needs Node.js 22+ and a headless Chromium)

1 skill

Still-image editing with ImageMagick

1 skill

Media offload from cards, recorders and URLs with checksum-verified copies, dual backup and a clip inventory

1 skill

Observation only: measures facts about a media file, including speech regions

1 skill

Titles and lower thirds burned into a programme, logos, PiP and chroma key, and placement of pre-rendered animated layers

1 skill

Timeline interchange with Premiere Pro and DaVinci Resolve: OTIO / FCP7 XML / EDL + SRT export, conform packages, round trip of the editor's changes

1 skill

Rule-based QC: PASS/WARN/FAIL/UNKNOWN judgement against the plan or spec

1 skill

Remotion motion graphics on real footage, timed from ffmpeg-skill measurements (word-timed caption effects, jump cuts, beat-synced motion, audio visualizers)

1 skill

Slide import, slide-change detection, slide text

1 skill

Thumbnail rendering; AI-generated background assets via FLUX.2 [klein] 4B (ADR-012)

1 skill

Speech to timestamped transcript (local faster-whisper)

1 skill

Installs the whole of Velisa (formerly AI Video Production OS): every Skill below except the optional disc-authoring kits, and the routing skill that picks the right one.

1 skill

Cut, trim, concat, fit, still overlay as part of an edit; computes retake and duplicate candidates

1 skill

Orchestrator: plans a production and calls the Skills below

1 skill