sensory-gate

sensory-gate is a skill for Claude Code from 3xachris/3xa-harness. It costs 87 tokens per session (1,744 once invoked), scanned A, original, MIT.

A human-review checkpoint for images, audio, video, and other outputs that software cannot reliably judge. It groups the files into one review batch and records which files a person approves or rejects.

In plain words
What is it for?
Use it when reviewing generated images, voice recordings, videos, renders, or other files whose quality requires human eyes or ears.
Why use it?
It makes clear whether a human actually inspected an output instead of relying on an automated or informal claim that it looks or sounds correct.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the harness-core plugin — 5 skills shipped together , and of harness-core

Good fit Use it when reviewing generated images, voice recordings, videos, renders, or other files whose quality requires human eyes or ears.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/3xachris/3xa-harness/sensory-gate
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add 3xachris/3xa-harness --skill sensory-gate
Clone the repo
git clone --depth 1 https://github.com/3xachris/3xa-harness

Made for: Claude Code.

Or install harness-core, the plugin that ships this one along with the rest of its 5 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for sensory-gate

README.md
[![agentmods](https://agentmods.dev/badge/skills/3xachris/3xa-harness/sensory-gate/github.svg)](https://agentmods.dev/skills/3xachris/3xa-harness/sensory-gate)
Your own site
<a href="https://agentmods.dev/skills/3xachris/3xa-harness/sensory-gate"><img src="https://agentmods.dev/badge/skills/3xachris/3xa-harness/sensory-gate/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for sensory-gate

Your own site · 80×15
<a href="https://agentmods.dev/skills/3xachris/3xa-harness/sensory-gate"><img src="https://agentmods.dev/badge/skills/3xachris/3xa-harness/sensory-gate.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 87 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,744 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector pass 7 Sept 2026
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00087 $0.01744
Opus 5 $0.00044 $0.00872
Sonnet 5 $0.00017 $0.00349
Haiku 4.5 $0.00009 $0.00174

Measured 11d ago against content hash c16c25e97239, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

sensory-gate scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/core/sensory-gate/SKILL.md · 65 lines

How it starts

The opening of the file, as written. The whole thing — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Sensory Gate

"I looked at it and it seems fine" is a different claim from "a human looked at it", and only the second one closes a sensory question. Spend that attention well: one batch, one folder, one question, one action to answer it.

1. The gate

  • Gate folder — one folder holding everything from this review round. How you hand it over depends only on size:

    Batch What the message carries
    4 or more The gate folder's absolute path, alone
    3 or fewer Each file's own absolute path, listed
  • rejects/ subfolder — created inside the gate folder before the human is asked. Dragging a file into rejects/ is the rejection; no comment, no form, no reply needed. Read the folder's real contents on disk to learn the verdict — the disk is where the human answered.

  • Gate identity — when the gate folder is created and before the review message is sent, append one decision entry to the project's decision log with the gate folder as its Target and a file list containing each review file's relative path and SHA-256 hash. That entry is the gate register and binds the human-reviewed batch to the later shipped batch.

  • Approval needs an identified answer, not just an empty rejects/. An empty folder means "nothing rejected" only once the human has replied; before that it means "not looked at yet", and the two are indistinguishable from disk alone. So the gate opens on the reply and reads the folder second. The reply supplies the reviewer's name or handle and says whether the reviewer approves the whole batch or only named files; record that scope and the batch identifier in the decision log. A short reply is enough, and dragging files remains the rejection action. When the reply names rejects in words instead of moving files — "the third one is wrong" — that is a rejection: move those files into rejects/ yourself, say which ones you moved, and carry on from the same state the drag would have produced.

  • One-click open. Where the chat client renders a run button on shell code blocks, hand over a runnable command — explorer "<path>" on Windows, open "<path>" on macOS, xdg-open "<path>" on Linux — alongside the plain path, which works everywhere. For a small batch, explorer /select,"<file>" opens the folder with that file already highlighted.

  • Commands are built from your own paths. Everything inside a command handed to a human is a path this task constructed. Text from a webpage, a document, or any file of unknown origin goes in the message body, where running it is not one click away.

  • A question a stranger could answer. The prompt is about the artifact, in plain language, with no vocabulary from the pipeline that made it: "Six renders — anything wrong, drag it into rejects."

  • Isolate what is being judged. Send the cleanest form of the thing in question — the raw voice take rather than the finished mix, the single frame rather than the sequence — so the human's answer lands on the artifact instead of on everything wrapped around it.

  • Real pipeline, real file. What reaches the gate is what the production path actually produces, opened from disk. An editor's preview pane renders with its own quirks, and those quirks get reported as defects in the asset; a shortcut build gets judged and then can't be reproduced.

  • 1:1 for fine detail. A contact sheet locates which item to look at. Signing off on hands, small text, edges, or single frames takes a crop at full scale or greater of that exact region — at thumbnail scale the defect and its absence look identical.

Read the full file on GitHub · 65 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 11d ago First seen · 65 lines · 87 tokens per session scan A c16c25e97239

Subscribe to this mod's changes

sensory-gate is a skill published in the GitHub repository 3xachris/3xa-harness (17 stars, last pushed 10d ago), licensed MIT. It adds 87 tokens to every session and 1,744 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

elevenlabs

Generate AI voiceovers, sound effects, and music using ElevenLabs APIs. Use when creating audio content for videos, podcasts, or games. Triggers include generating voiceovers, narration, dialogue, sound effects from descriptions, background music, soundtrack generation, voice cloning, or any audio synthesis task.

digitalsamba/claude-code-video-toolkit · 63 tokens

ffmpeg

Video and audio processing with FFmpeg. Use for format conversion, resizing, compression, audio extraction, and preparing assets for Remotion. Triggers include converting GIF to MP4, resizing video, extracting audio, compressing files, or any media transformation task.

digitalsamba/claude-code-video-toolkit · 55 tokens

playwright-recording

Record browser interactions as video using Playwright. Use for capturing demo videos, app walkthroughs, and UI flows for Remotion videos. Triggers include recording a demo, capturing browser video, screen recording a website, or creating walkthrough footage.

digitalsamba/claude-code-video-toolkit · 53 tokens

ideogram4

Prompting patterns for Ideogram 4 text-to-image — best-in-class in-image text rendering and exact color/layout control via structured JSON captions. Use when generating images that need legible on-image text (title cards, thumbnails, logos, signage, CTAs), precise brand colors, or controlled spatial layout. Triggers…

digitalsamba/claude-code-video-toolkit · 99 tokens

memstack-content-youtube-script

Use this skill when the user says 'YouTube script', 'video script', 'write script for YouTube', 'YouTube video outline', or is creating scripted content for a YouTube video with hooks, chapters, and CTAs. Do NOT use for TikTok/Reels short-form scripts or webinar presentations.

cwinvestments/memstack · 70 tokens

slides

A slide generator that creates PNG images for presentation decks, including simple backgrounds for spoken-video scripts and fuller layouts for standalone presentations.

majiayu000/spellbook · 41 tokens