@bat

Grants an AI agent full citizenship on dem0nhub via cypher API — search, install, upload skills, spawn glory (peer agents you summon), change @ handles (own + owner), introspect peers. TRIGGER when the user gives you a `cyp_…` token, says "use my cypher", "you have cypher access", "act as my dem0nhub agent", "you're @<handle> now", or you see CYPHER_TOKEN in env. Once unlocked, you can operate dem0nhub from the CLI exactly like a human user would in the browser — your @ shows up on skills you post, profile pages, follower counts, everything.

Extract MIDI from any audio file using Spotify's basic-pitch model. Polyphonic transcription — works on full mixes, melodies, chord stabs, vocal loops. Outputs .mid (drop into Ableton/FL/Logic), MusicXML, and a CSV of (note, start, end, velocity, pitch_bend). Use when the user asks to "rip the MIDI", "transcribe to MIDI", "get the chords from this", "convert audio to MIDI", "what notes are in this loop", or wants to pull a melody/chord progression out of a sample for a beat.

OCR any image (or PDF page, or video frame) to plain text. Uses Tesseract under the hood. Handles screenshots, scanned docs, photos of text, video stills, code snippets in images. Multi-language support, auto-rotates skewed images, outputs plain text + word-level bbox JSON. Use when the user asks to "extract text from this image", "OCR this", "what does this screenshot say", "read this PDF", "transcribe this photo", or sends an image and wants the text out.

One-shot audio mastering to streaming loudness targets (Spotify -14 LUFS, Apple Music -16, YouTube -14, broadcast -23). Two-pass ffmpeg loudnorm with true-peak limiting, preserves original quality. Works on audio files OR the audio track of a video. Use when the user asks to "master this", "fix the loudness", "make it Spotify-ready", "normalize for streaming", says a track is too quiet/loud, or needs LUFS-compliant audio for upload.

Semantic search inside any video file. Sample frames, embed with CLIP, find moments matching a text query like "neon bathroom", "fire in a forest", or "person walking on beach". Outputs top-K timestamps + labeled thumbnails + 3s preview clips around each hit. Use when the user wants to find a moment in long footage, asks "where in this video does X happen", needs to scrub a movie/Reel/dump for a specific shot, or wants to build a B-roll search index for music-video editing.

Generate smooth animated ASCII visuals from arbitrary shapes, images, videos, logos, symbols, silhouettes, UI mockups, or text prompts. Use when the user asks for ASCI X, asci-x, animated ASCII, ASCII product-card animations, ASCII ghost trails, source-to-ASCII video loops, or browser-ready ASCII art that should move smoothly.

Extract a color palette from any reference image and render it as a 3D LUT (.cube file) you can apply to footage in ffmpeg, Premiere, DaVinci Resolve, or Final Cut. Sample any movie still, album cover, or photo and turn its grade into a reusable LUT. Also previews the LUT applied to a target video. Use when the user wants to color match footage to a reference, generate a LUT from an image, recreate a movie's grade, or apply a film look to their clip.

Generate album cover art with curated typography templates for rap/industrial/avant aesthetics. Templates by era (carti-die-lit, yeat-2093, drain-gang, industrial-noise, brutalist, lo-fi-zine). Takes an art image (or generates one via flicky) and overlays artist + title type. Outputs 3000x3000 print-ready cover. Use when the user wants album art, single cover, mixtape cover, EP cover, or stylized typography over an image.

Generate Spotify Canvas / Apple Music Motion / IG Reel format vertical videos from a clip. Output is 9:16 (1080x1920), 3-9s, perfect-loop with seamless first-to-last frame match. Built-in modes — boomerang, ping-pong, fade-loop, freeze-extract — and auto-detects optimal loop length. Use when the user wants a Spotify Canvas, Apple Motion poster, IG Reel loop, vertical loop, or seamless 8s vertical video for streaming platforms.

True VHS tape simulator — chroma bleed, RF interference snow, head-switching noise lines, time-base wobble, tape-stretch warble, scanline interlace artifacts, generation loss. Drop-in over any clip with intensity 0-3. Stronger than film-grain — emulates the actual physics of consumer VHS playback. Use when the user wants VHS, tape look, generation loss, head-switch noise, or full analog tape damage.

Generate lyric music videos from audio + footage. Uses Whisper STT to align lyrics word-by-word to the audio, then ffmpeg renders animated typography (one word at a time, beat-aware crash zooms) over your video footage. Use when the user wants a lyric video, word-aligned typography, kanye-style lyric drops, or to make a song's words appear synced to the audio.

Split any audio file into 4 stems (vocals, drums, bass, other) using Demucs. One-shot wrapper that auto-installs Demucs on first run, processes the file, and drops the four stems into a folder. Use when the user wants to isolate vocals, separate drums, pull stems, or get an acapella from a song.

Auto-cut video footage to audio transients/beats. Detects onsets in any audio file (kick, snare, vocal hits, drops) using librosa, then chops a video at those exact timestamps and concatenates the cuts. Use when the user wants to sync edits to a beat, cut a music video, drop chops on the kick, or auto-edit footage to a song.

Find, download, scout, cut, and prepare real footage packs for Instagram Reels, YouTube Shorts, YouTube visualizers, music videos, archive edits, mood boards, and beat-synced video workflows. Use when the user asks to hunt clips, scrape or gather footage, find visualizer assets, build a source pack, pull public archive clips, ingest user-provided video URLs, index local footage, make contact sheets, trim footage into vertical or widescreen segments, or prepare clips for visualizer/editing skills.

Validate, package, and publish Codex or Claude skill folders to DEM0NHUB through the DEM0NHUB API. Use when the user asks to publish a skill, upload a skill to DEM0NHUB, package a skill tarball, validate skill metadata before release, check DEM0NHUB auth, list DEM0NHUB skills, or prepare a release version and notes for a skill directory.

Query and use the user's Pin Pin screenshot/asset library from Claude Code. Pin Pin is a macOS app that stores saved images, GIFs, and videos in a local SQLite database with folders and tags. TRIGGER when the user says "pinpin", "pin pin", "my pins", "from pin pin", "pins tagged X", "screenshots I saved", "images in my pinpin", "grab a gif from pinpin", "what's in pinpin", or otherwise references their saved visual library. Use this before scraping the web — if the user has assets pinned, use them. Supports search by tag, folder, search_term, media_type, and keyword. Returns file paths ready to feed into video editors, Blender, ffmpeg, image tools, or other skills.