maddexritter-rgb

21 mods across 1 repository, 3 stars between them.

maddexritter-rgb/vibe-editing

Instructions file

Instructions for maddexritter-rgb/vibe-editing, covering vibe editing — how to run this project, making clips (the main job), horizontal "mid" videos (the highlight skill), first-time setup (only if it isn't set up yet) and brand it / change it.

3 2mo ago A 838 tokens original MIT

vibe-editing

03

maddexritter-rgb/vibe-editing

Plugin Claude Code

Raw footage in, finished captioned vertical clips out — with one command. Ingests a local file, a URL, or a cloud-drive link; transcribes; mines the best moments with a real editorial rubric; hand-cuts; face-tracks to 9:16; burns word-level captions; mixes music; renders (manifest-driven + stage-cached); and runs a…

3 2mo ago A tokens not measured original MIT

audit-audio

04

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Dedicated audio auditor. Receives a rendered clip and checks ONLY audio quality: word clipping, pops/clicks at splice points, level consistency, buzz/hum, music bed balance, clean open/close. Runs in fresh context — audio only, no video. Returns structured pass/fail per check with timestamps for failures. Part of the…

3 2mo ago A 83 tokens original MIT

audit-captions

05

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Dedicated caption auditor. Receives a rendered clip and checks ONLY caption quality: accuracy vs spoken audio, speaker color attribution, timing sync, formatting, emphasis, and positioning. Runs in fresh context with NO knowledge of how the clip was made. Returns structured pass/fail per check with specific timestamps…

3 2mo ago A 80 tokens original MIT

audit-script

06

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Dedicated editorial/script auditor. Receives ONLY the transcript text of a rendered clip (no video, no audio). Checks logical flow, context-before-payoff, cold viewer comprehension, one-arc structure, hook quality, brand safety, and SOP compliance. Uses all editorial SOPs (post bible, scorecard, Q&A SOP, hooks…

3 2mo ago A 118 tokens original MIT

audit-visual

07

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Dedicated visual auditor. Receives a rendered clip with audio STRIPPED and checks ONLY visual quality: face tracking stability, framing consistency, jump cuts, frozen frames, black frames, rendering artifacts. Cannot see captions (they're burnt in — use audit-captions for that). Runs in fresh context with NO editing…

3 2mo ago A 96 tokens original MIT

caption-clips

08

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

THE single caption skill. There is ONE caption style for ALL footage — "spice" (the locked the reference editor look). Burns it onto any short-form clip via ONE engine. Default and only style. Use whenever the user says "caption this", "add captions", "pro/speaker/spice/montserrat captions", "caption these clips"…

3 2mo ago A 139 tokens original MIT

edit

09

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

THE master editing skill. Raw footage → finished clips. ONE spine for ALL footage — monologue, Q&A, hotline, podcast, multicam, any of it. Auto-detects speaker count (1 or 2) and camera count (1 or multi). This skill ORCHESTRATES — it owns mining and boundary-picking, then POINTS TO the standalone skills for each…

3 2mo ago A 286 tokens original MIT

footage-fetch

10

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Download raw footage FAST from Google Drive (folders OR single-file links) into properly-scaffolded per-download projects, with integrity verification. Use whenever Operator drops a Google Drive link and wants the footage pulled to disk — "download this footage", "grab this drive folder", "pull these clips", "add this…

3 2mo ago A 137 tokens original MIT

highlight

11

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

THE highlights-channel clipper, run three ways. MINE mode (given a link or file of a raw Q&A / hotline / podcast recording) mines it into RANKED, postable HORIZONTAL 16:9 "mid" videos optimized for SUBSCRIBER growth (subsper1kviews, NOT views/CTR), cuts them, optionally appends YOUR own CTA outro, and delivers them…

3 2mo ago A 271 tokens original MIT

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

THE single face-tracking + 16:9→9:16 reframe skill (locked house style, 2026-06-10). Two cases — that's it. (1) SINGLE FACE → scripts/qareframev2.py --preset {talking-head | stage | split-top | guest | podcast} — Y-LOCK (eyeline fixed at clip median, only X keyframes) + face-BOX center (steadier than nose tip). Named…

3 2mo ago A 394 tokens original MIT

listicle-short

13

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

One-shot pipeline that turns a long-form video (YouTube link or local file) into a finished 60-second NUMBERED rapid-fire "listicle" short in the locked Speaker/SF style — the speaker's punchiest verbatim soundbites, cut + assembled, face-tracked 9:16, SPICE captions (via the caption-clips skill) with persistent.

3 2mo ago A 82 tokens original MIT

long-form-ingest

14

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Transcribe a long-form video to word-level timestamped JSON and detect scene boundaries. Use whenever a new video file enters the pipeline and a transcript does not yet exist. Produces transcript.json (word-level with timestamps), scenes.json (visual cut points), and meta.json (duration, fps, resolution). Every other…

3 2mo ago A 75 tokens original MIT

promo

15

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

THE branded video-promo skill (run as /promo). Makes premium branded videos with Remotion + Claude Code, picking ONE OF TWO MODES at the start. MODE A — ANIMATED EXPLAINER / DEMO — fully animated, brand-tokenized motion graphics (the LIQUID-GLASS kit in templates/glass: brand sting, hook cards, content-flywheel ring…

3 2mo ago A 314 tokens original MIT

render

16

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

THE manifest-driven, stage-cached rendering engine. Every clip is a recipe (manifest.json + source files for each layer — cuts.json, reframe.json, captions.ass, music.json); render builds it in ordered stages (cut → reframe → grade → captions → mix → leadfix → deliver); each stage's output is content-hashed into…

3 2mo ago A 290 tokens original MIT

scorecard-audit

17

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Runs the company.com 51-rule scorecard against a finished short. Three format variants auto-selected: qa (151 max / pass=121), hotline (121 max / pass=97), story (145 max). Catches what sf-audit can't — narrative coherence (N1-N10 including the most-failed rule N4 "payoff is the final line"), hook structure (H1…

3 2mo ago A 199 tokens original MIT

script-cut

18

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Text-based / script-driven precision cutter for talking-head clips. Use this whenever you need to cut a clip from a transcript and the recurring problems are clipped words, leftover "uh"/"um", false starts or restarts, dead air, scrambled word order, or a payoff that gets chopped. Instead of trusting the editor's…

3 2mo ago A 165 tokens original MIT

sf-audit

19

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

The final-review checklist every short-form edit ships against. Based on the Team Speaker V1 MVP Standards (the studio). 16 checks (updated 2026-05-04 from 13 per the reference editor feedback), grouped into Subtitles / Audio / Video. Automates 12 of them fully, flags the rest for a manual eye.

3 2mo ago A 173 tokens original MIT

source-intel

20

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Footage intelligence agent. Analyzes raw source footage BEFORE any editing begins. Produces a structured report: face positions/sizes, camera angles, scene changes, audio quality, speaker count, movement patterns. Feeds into edit step 1 (DETECT) and informs every downstream decision (reframe zoom, caption height, mic…

3 2mo ago A 90 tokens original MIT

watch

21

maddexritter-rgb/vibe-editing

Skill Claude CodeCodex

Give Claude real eyes + ears on video. Two jobs: (1) WATCH / COMPARE — understand what is in a clip, and compare it against a reference reel or a previous version; (2) QA-EYES — read burnt-in captions (OCR), check caption placement, framing, pacing and loudness on a finished 9:16 short. Wraps the claude-video-vision…

3 2mo ago B 196 tokens original MIT