Hz Quest Verify First logo

Hz Quest Verify First

Organization
meta-quest
hz-quest-verify-first

MANDATORY pre-flight check before answering any question or writing any code related to Meta Quest VR headsets (Quest 2, Quest 3, Quest 3S, Quest Pro) or apps that target them. Forces verification against authoritative Meta sources via the metavr CLI / metavr MCP tools BEFORE relying on training-data knowledge. Counters the failure mode where agents answer Quest-related questions from stale memory and ship deprecated APIs, broken Android manifests, and store-rejected builds. Loads automatically when the user is in a Quest project (any reference to Oculus / Meta Quest / Horizon OS, Unity OVR or Meta XR packages, com.meta.* / com.oculus.* package IDs, Quest-targeted AndroidManifest, .meta files, or developers.meta.com / developer.oculus.com URLs).

Overview

Publishermeta-quest
Repositoryagentic-tools
Skill namehz-quest-verify-first
Stars
195
Forks
17
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by meta-quest on GitHub. Read the source before you install it.

Installation

Install the Hz Quest Verify First AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/meta-quest/agentic-tools.git /tmp/agentic-tools
mkdir -p .claude/skills
cp -r /tmp/agentic-tools/skills/hz-quest-verify-first .claude/skills/hz-quest-verify-first
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Hz Quest Verify First in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Hz Quest Verify First on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Hz Quest Verify First is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Verify First — Meta Quest Development

What you need to know before reading the rest of this

You are about to be told to call tools you may not recognize by name. Read this section first so the rest makes sense.

  • Meta Quest is Meta's line of standalone VR / mixed-reality headsets (Quest 2, Quest 3, Quest 3S, Quest Pro). They run an Android-based OS.
  • Horizon OS is the name of that operating system. It is forked from Android, ships on every Quest headset, and Meta updates it on its own cadence (separate from upstream Android). Many APIs are Horizon-OS-specific and have no analog in stock Android.
  • metavr ("Meta VR CLI") is a Rust CLI tool published by Meta as metavr on npm. Invoke via npx -y metavr <args> — no install required; npx always pulls the latest published version. It wraps adb and Meta's developer APIs into higher-level commands for Quest device development: list devices, install apps, capture Perfetto traces, search Meta Quest documentation, query 3D asset libraries, automate UI interactions, etc. metavr is the primary action layer for Quest dev work — the equivalent of what gcloud is for GCP or git is for source control.
  • metavr MCP server is a built-in mode of metavr that exposes a focused set of tools to AI coding agents over the Model Context Protocol. The relevant tools for this skill are:
    • meta_docs_search — search the official Meta developer documentation
    • meta_docs_get_page — fetch the full text of a specific docs page
    • device — query and control connected Meta Quest headsets (list, info, connect, reboot, battery, controllers, proximity, etc.)
    • app — query and manage installed apps (list, info, install, uninstall, launch, stop, clear)
    • files — file ops on a connected headset (ls, push, pull, rm, mkdir)
    • run — catch-all for any metavr subcommand without a dedicated tool (perf, ovrmetrics, ui, audio, casting, window, unity, sideload, asset, config, …). Its JSON Schema is generated from clap so the available subcommands and their typed args are visible to you in the tool definition.
    • cli_help — discover metavr subcommands and flags as markdown
  • Meta SDKs you may not recognize that are commonly used in Quest projects: Meta XR All-in-One SDK (Unity), Meta XR Core / Interaction / Platform / Voice / Movement SDKs, Meta Spatial SDK (Kotlin / Android panels), IWSDK (Immersive Web SDK for WebXR), MRUK (Mixed Reality Utility Kit), Presence Platform, Horizon Platform SDK. These ship on their own release cadences, separate from Quest OS.
  • OVR vs Meta XR — older Quest Unity content and most of the public web references the OVR namespace and the "Oculus Integration" Unity package. Both have been deprecated in favor of the Meta XR All-in-One SDK family. Most training data still recommends the deprecated names. Verify before recommending anything in this area.

Why this skill exists

Meta Quest and Horizon OS evolve fast. Meta SDKs ship every few weeks. APIs get renamed before public release. Manifest entitlements and store policies update monthly. Entire SDK families (Meta Spatial SDK, IWSDK, Horizon OS shells) post-date most LLM training corpora.

You almost certainly have stale knowledge about Quest specifics. The package names you remember have been renamed. The Unity menu paths have moved. The setup steps that worked on a prior Horizon OS version silently broke on the current one. The "OVR" namespace you reach for has been split across half a dozen Meta XR packages.

If you answer a Meta Quest question or write Quest-targeted code without verifying first, you will hallucinate. The user will not notice until their build fails, their app gets rejected from the store, or their AndroidManifest doesn't publish.

This skill is a hard gate. Treat it as a precondition, not a suggestion.

When this skill applies

You MUST run the verification flow below if ANY of the following is true for the current task:

  • The user mentions Meta Quest, Quest 2 / 3 / 3S / Pro, Oculus, Horizon OS, MR, VR, XR, OpenXR (on Quest), Meta SDK, Meta XR, OVR, Spatial SDK, IWSDK, MRUK, Movement SDK, Voice SDK, Presence Platform, Horizon Platform SDK, hand tracking, passthrough, anchors, scene mesh, casting, eye tracking, body tracking, controllers, or platform compliance / store policy
  • The repository contains any of:
    • Unity files with OVRPlugin, "Meta XR All-in-One SDK", or com.meta.xr.* in manifest.json / Packages/manifest.json
    • Spatial SDK gradle plugin (com.meta.spatial.plugins.*)
    • AndroidManifest with com.oculus.*, com.meta.quest.*, or horizon feature declarations
    • Package IDs starting with com.meta.* or com.oculus.*
    • .meta files indicating Unity, or Quest_*.uplugin / *.unrealhxr files
  • The user's question references developers.meta.com/horizon, developer.oculus.com, oculus.com/sparkle-updates, or any Meta Quest documentation URL
  • The agent is about to recommend any Meta SDK API, namespace, package, or AndroidManifest entitlement
  • The agent is about to answer a question about app review, Meta Horizon Store submission, store-readiness or publishing requirements, content rating, or distribution
  • The agent is about to write adb commands targeting a specific device
  • The agent is about to claim what is or isn't installed on the user's headset

If you are not sure, the answer is YES — run the verification flow.

The verification flow

Step 1 — Verify against authoritative documentation

Before writing or recommending anything Quest-specific, call the meta_docs_search MCP tool. If MCP is not available, use the equivalent metavr CLI command metavr docs search.

MCP:

meta_docs_search(
  query="<the specific claim or API you are about to make>",
  scope="auto",   # or unity / unreal / spatial_sdk / android / native / web /
                  #    policy / distribution / design
  mode="verify"
)

CLI:

bash
metavr docs search "<query>"

If you need exact wording (manifest entries, full API signatures, store policy text, code snippets), follow up with meta_docs_get_page on the canonical_url or doc_path returned by verify. Never paraphrase a truncated snippet when correctness matters.

CLI:

bash
metavr docs fetch "<canonical_url_or_path>"

Step 2 — Verify the user's actual environment

Before suggesting which device a command should target, claiming an app is installed, recommending an adb command, or writing install / launch / sideload steps, query the user's actual metavr-managed environment.

MCP:

device(action="list")                        # ALWAYS start here
device(action="info", target="<serial>")
app(action="list", target="<serial>")
app(action="info", package="<package>")
run(subcommand=["config", "show"])           # catch-all for misc reads

CLI:

bash
metavr device list
metavr device info <serial>
metavr app list -d <serial>
metavr app info <package>
metavr config list

The user may have zero, one, or many headsets connected via USB and WiFi — multiple Quest models, dev kits, sideloaded builds, pinned older Horizon OS versions. Your training data has zero visibility into this.

Step 3 — Discover metavr capabilities when unsure

If you do not know which metavr subcommand or run invocation fits the user's request, call cli_help (MCP) or metavr --markdown-help (CLI) first. Do not invent flags or subcommands — metavr gets new functionality every release and your training data does not include it.

For long-tail subcommands (perf, ovrmetrics, ui, etc.), the run tool's input schema enumerates every subcommand path and its typed args via JSON Schema oneOf. You can read it directly from the tool definition rather than guessing.

MCP:

cli_help(topic="perf")   # focused help for a subcommand tree
cli_help()               # full top-level command tree

CLI:

bash
metavr --markdown-help
metavr perf --help

Specific failure modes you cause by skipping verification

These are concrete, recurring failures that the verification flow prevents:

  • Deprecated APIs that compile but no-op at runtime. OVR-namespace function names that were forwarded for one release and removed the next. Code looks correct, builds clean, runs without errors — and produces no observable behavior on device.
  • AndroidManifest entitlements that get the app rejected from the Meta Horizon Store. The required entitlement names changed; an entitlement you "know" is needed is no longer recognized; a new mandatory feature flag is missing.
  • Meta SDK package names that don't exist. Meta XR All-in-One SDK is a family of UPM packages with version-specific names; guessing produces "package not found" errors.
  • Unity menu paths from outdated tutorials. Meta XR menu items moved between SDK versions; setup steps that reference Oculus → ... may now live under Meta XR → ... or have been removed entirely.
  • Spatial SDK class names that were renamed before public release. The Spatial SDK iterated on its public API late; many class and method names in older blog posts or training data are wrong.
  • Passthrough / anchor / scene-mesh setup that worked on a prior Horizon OS version. Permission flow, system overlay APIs, and manifest features changed across Quest OS versions. What worked in a tutorial last year may silently fail on current Horizon OS.
  • OpenXR extensions that are listed but not supported on Quest's runtime. Generic OpenXR documentation lists extensions Meta has not implemented. Always verify against Meta's documentation, not the OpenXR spec.
  • Wrong device targeted. Recommending a adb shell command without first listing devices, then watching it fail or hit the wrong headset.
  • Store policy claims that are out of date. Review requirements, rating buckets, and distribution rules update — verify before answering user questions about submission.

Anti-patterns

Do not do any of these:

  • Answer a Meta Quest question without calling meta_docs_search first because "you remember" the answer
  • Recommend an adb shell command without first calling device(action="list") to see what's connected
  • Guess a package name, namespace, or class name from training data when a verify call would resolve the ambiguity
  • Paraphrase a verify-result snippet for a manifest entry, API signature, or store policy when exact wording is required (use meta_docs_get_page)
  • Search the open web for Meta Quest documentation when authoritative Meta docs are one tool call away
  • Skip verification because "the docs probably say X" — verify, then say X
  • Assume metavr does not have a subcommand for what the user wants — check cli_help first

Tone and presentation

When you do verify, present your answer with citations and a brief recency note. The user is paying for the verification step — show that it happened.

Per the current Meta Spatial SDK docs (verified just now via
meta_docs_search):

  <answer grounded in retrieved content>

Source: <canonical_url returned by the tool>

This both proves the verification ran and trains the user to trust verified-answer responses over from-memory ones.

If the metavr MCP tools are not available

If the agent host does not have the metavr MCP server installed, fall back to the metavr CLI invoked via npx (same authoritative backend, same content) and tell the user once how to install the MCP server for next time:

bash
npx -y metavr mcp install <your-tool>   # claude-code | cursor | claude-desktop | vscode | ...
npx -y metavr mcp install project       # install at the repo root for this project only

The verify step is required, not optional — run it via npx -y metavr docs search "<query>" even on the very first question if the MCP server is not yet wired up.

Related skills

  • metavr-cli — full metavr CLI reference (commands, flags, examples)
  • hz-vr-debug — on-device debugging with logs and screenshots
  • hz-perfetto-debug — Perfetto trace analysis for jank / GPU / CPU bottlenecks
  • hz-store-submit — end-to-end Meta Horizon Store submission (includes store-readiness checks)
  • hz-spatial-sdk — Meta Spatial SDK API guidance
  • hz-platform-sdk — Horizon Platform SDK API guidance

Frequently asked questions

What does the Hz Quest Verify First AI skill do?

MANDATORY pre-flight check before answering any question or writing any code related to Meta Quest VR headsets (Quest 2, Quest 3, Quest 3S, Quest Pro) or apps that target them. Forces verification against authoritative Meta sources via the metavr CLI / metavr MCP tools BEFORE relying on training-data knowledge. Counters the failure mode where agents answer Quest-related questions from stale memory and ship deprecated APIs, broken Android manifests, and store-rejected builds. Loads automatically when the user is in a Quest project (any reference to Oculus / Meta Quest / Horizon OS, Unity OVR...

Why use Hz Quest Verify First on TypingMind?

Because you install it once and use it with any model. Hz Quest Verify First is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Hz Quest Verify First in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/meta-quest/agentic-tools/tree/main/skills/hz-quest-verify-first. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Hz Quest Verify First?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Hz Quest Verify First?

As many as you like. As long as a model supports skills, you can use Hz Quest Verify First with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Hz Quest Verify First AI skill free?

Yes. It is published on GitHub by meta-quest under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇