Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    AI Video Publication QC Gate — Per-Video Checks

    1

    Never publish an AI video you haven't actually watched — and never spend render credits on the wrong person.

    $39

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    AI Video Publication QC Gate — Per-Video Checks

    AI Video Publication QC Gate — Per-Video Checks

    Example session with this skill installed

    Run Phase 3 QC on video_082.mp4 against the manifest. Presenter is using the 'sophia_canonical' reference. Check for identity drift, lip-sync, and product stability.

    • Read your context and instructions
    • Compiled the ai video publication

    I have completed the Phase 3 technical audit for video_082.mp4. The video was compared against the sophia_canonical reference set. While the start image passed Phase 1, the full-track review revealed a failure in product fidelity and text stability.

    QC Verdict: video_082.mp4

    CheckStatusNotes
    Identity (Face/Hair)PASSMatches 'sophia_canonical' hair texture and facial structure.
    Body / OutfitEXEMPTPresenter is wearing a heavy knit sweater; build check skipped.
    HandsPASSTwo distinct hands visible; no extra digits or blending.
    Text StabilityFAILProduct label text flickers and morphs at 0:04 mark.
    Product FidelityFAILThe 'GlowSerum' bottle cap changes from gold to silver at 0:06.
    Voice / Lip-SyncPASSFull track sync is accurate across all 12 seconds.
    Duration MatchPASSVideo (12.2s) matches narration (11.8s) within tolerance.
    LoudnessPASSIntegrated loudness measured at -23.1 LUFS.
    Hook / End-CardPASSWording is within 90% frame width; no cut-offs.

    Verdict: FAIL
    Action: The item has been marked as FAIL in the manifest. Publishing is blocked.

    Next steps
    1.

    Re-roll Clip: Regenerate the middle segment with a specific prompt lock for "gold bottle cap" and "stable label text".
    2. Update Manifest: A new QC run is required after the file swap to clear the gate.
    3.

    Review Prompt: Ensure Phase 2 prompt wording explicitly defines the product cap color to prevent the silver morph.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    AI video fails expensively and quietly: the start image is subtly the wrong person, the outfit changes color between clips, the product morphs in the presenter's hand, three sampled frames look perfect while the full track drifts and drops audio. Batch-level "approved" flags hide all of it.

    This skill runs a per-video, per-check gate in three phases:

    • Before spending: verify the start image against a canonical reference set (portrait, face-cropped still, one approved video) — face and hair checked as identity, verdict recorded before any video credits are spent
    • In the prompt: outfit locked word-for-word across clips, state-sensitive product states called out explicitly, product-in-hand from frame one
    • On the finished file: a full-duration watch covering identity drift, hands, product fidelity frame-by-frame against the source listing, text flicker/morph, real-motion checks, voice and lip-sync judged on the complete track, video-vs-narration duration match, loudness measurement, and hook/end-card framing

    Plus the publishing rules that make the gate real: per-video QC records in the manifest are the only publishing authority, failed checks are recorded with reasons so repairs re-roll one clip instead of a batch, and QC re-runs after every repair or file swap.

    Who it's for: AI UGC creators, avatar/presenter video producers, faceless-channel and product-demo studios, and agencies shipping AI video for clients. What the buyer gets: one self-contained SKILL.md — setup, all three phases, every check defined, verdict-recording format, and engine/voice change protocol.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Video Skills.

    Requires a pre-defined canonical reference set. Identity checks for body build are skipped for baggy outfits. Requires full-track duration analysis rather than sampling.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 8 days ago

    Permissions

    No special permissions declared or detected

    Listed8 days ago

    What's inside

    Frequently Asked Questions

    This skill implements a three-phase gate that verifies identity and product fidelity before and after rendering, preventing expensive credit waste on videos with identity drift, morphing products, or lip-sync errors.

    The skill is engine-agnostic and works with any manifest-driven workflow or pipeline, including HeyGen, Sora, Runway, Kling, and Luma.

    You will receive the complete logic for the three-phase validation workflow, integration schemas for reference assembly, and the atomic verdict logging structure for your records.

    Setup requires providing a canonical reference set, including a portrait and a benchmark approved video, to serve as the 'source of truth' for the identity and product audits.

    While it tracks facial features and product stability with high precision, it cannot currently audit body build consistency for presenters wearing baggy clothing or thick sweaters due to lack of visibility.

    Unlike simple prompting, this skill performs frame-by-frame temporal analysis and technical validation of audio-visual sync, catching artifacts that mid-shot 'hallucinations' often create.

    New in Image & Video