SkillAgentSearch skills...

masked-v2v-audit

Upstream data-dependency, statistical outlier, frame-alignment, and visual verification workflow for H3 masked V2V pipelines. Invoke when auditing upstream inputs before queuing renders or remakes, or when debugging unexpected behavior across multi-stage video pipelines.

Install / Use

npx skills add fblissjr/ComfyUI-h3-explorations --skill masked-v2v-audit

Installs into whichever agent you are using.

About this skill
♊

Gemini Rules

Gemini CLI config

Quality Score

60/100

Category

Automation

Supported Platforms

Gemini CLI

Our assessment of masked-v2v-audit

masked-v2v-audit scores 60/100 on our quality scale, 2682nd of 2,790 Automation skills we index.

Its Gemini Rules is 3.5 KB long, lightly structured (2 headings) and no code examples: a solid amount of guidance for an agent.

It has no GitHub stars yet, so there is no community track record; judge it on its content.

Substance
26/30
Structure
5/20
Description
15/15
Adoption
0/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated 2 days ago, so masked-v2v-audit is actively maintained.
  • No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
  • Its trust signals score 80/100, with 2 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-11-14. Automated pattern scan on 2026-11-14. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

masked-v2v-audit compared with similar skills

All 4 of these similar skills score higher than masked-v2v-audit; compare them before choosing.

SkillScoreStarsUpdatedFormat
masked-v2v-audit (this skill)by fblissjr6002d agoGemini Rules
Agent-Reachby Panniantong100112.3k8d agoCLAUDE.md
rufloby ruvnet10081.9ktodayCLAUDE.md
Claude-Code-Game-Studiosby Donchitos10039.7k7d agoCLAUDE.md
Scraplingby D4Vinci100103.6k2d agoMCP Server

Frequently asked questions

How do I install masked-v2v-audit?
Run npx skills add fblissjr/ComfyUI-h3-explorations. The install tabs above show the steps for each supported agent.
Which AI agents does masked-v2v-audit work with?
It is written for Gemini CLI, as a Gemini Rules file. Other agents that read the same format can often use it too.
Is masked-v2v-audit safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It declares no license and scores 80/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is masked-v2v-audit still maintained?
The repository was last updated 2 days ago, so masked-v2v-audit is actively maintained.

name: masked-v2v-audit description: Upstream data-dependency, statistical outlier, frame-alignment, and visual verification workflow for H3 masked V2V pipelines. Invoke when auditing upstream inputs before queuing renders or remakes, or when debugging unexpected behavior across multi-stage video pipelines. reviewed: 82753cd8

Masked V2V Upstream & Pre-Render Diagnostic Skill

Passing an automated preflight gate (0 likely to fail) is necessary but not sufficient. Whenever working on masked V2V loads, apply the general diagnostic principles below and consult docs/wiki/masked_v2v/upstream_anomaly_and_alignment_checks.md and docs/wiki/masked_v2v/chain_trace_and_remakes.md §1 for the stage-by-stage reprocessing DAG.

Core Diagnostic Habits (Always Apply)

  1. Trace Every Explicit and Hidden Dependency in the DAG (Never Assume Inputs Are Synced):
    • Audit every model stream and JSON/NPZ table feeding the stage — not just the primary track mask. In H3 masked V2V, multiple independent models feed the same stage without automatic cross-locking (e.g., 3D SAM3DBody _mesh vs. 2D Sapiens2-Pose _points.npz vs. _pose_table.json in head_reference.py; MoGe-2 depth.npz in shot_to_load.py job; seen_looks in shots.json).
    • Check whether upstream diagnostic tables (marks_against_the_body_model, face_score_mean, part_grades.json, seen_looks) are actually acted on by the code or merely logged as "decides: nothing". If the code logs an anomaly without acting on it, you must pass the corresponding override/exclusion flags (--no-head-on, --no-face-on, --no-draw-face-on, --drop, --depth) or fix the stage writer and reprocess downstream.
    • Verify that every input video, mask, and array feeding a job has the exact same start frame, frame count, and source cut.
  2. Interrogate Summary Statistics Against Per-Frame Reality (Least / Median / Most):
    • Whenever any stage prints summary metrics (least / median / most, whole-shot bounding box vs. per-frame box, area steps, or score floors), compare the extremes and whole-shot aggregates against the median frame.
    • A large gap between the aggregate/extremes and the median means a small number of outlier frames (occlusions, motion blur, tracker jumps) are distorting whole-shot behavior. Inspect those frames directly on disk.
  3. Visually Audit Intermediate Conditioning Against the Source Film Using Native read_file First:
    • Never treat intermediate stages (masks, keep-outs, 3D mesh renders, pose landmarks, crop boxes, or generated prompt text) as black boxes.
    • Use agy's built-in read_file (view_file) tool directly on the pre-built .1fps.<first>-<last>.mp4 look strips (first and on-plan), <Video 1> reference .mp4 and _sheet.jpg, and bench/inspect_mask_steps.py --strips <out.jpg> JPEGs first, and extract custom side-by-side contact sheets when zooming into transitions, rapid motion, occlusions, or flagged spans.
  4. Verify Landed Results With Native read_file on .1fps.mp4 & 24 fps .mp4 (Plus Close-Up Sheets Where Needed) Before Paging:
    • After rendering and landing, inspect the landed <load>_<first>_<last>_stacked.1fps.<first>-<last>.mp4 strip and 24 fps .mp4 clip directly via read_file (plus close-up contact sheets comparing ORIG, prior attempt, and CUT on targeted frames) before paging to the review server.

Related Skills

View on GitHub
GitHub Stars0
CategoryAutomation
Updated2d ago
Forks0

Trust signals

80/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

1 medium1 low
masked-v2v-audit — Gemini Rules: Install & Safety Check | SkillAgent