一键在 Manus 中运行任何 Skill

quickstart

星标9

分支1

更新时间2026年7月8日 21:29

Get a first-time user to a real 3D asset in seconds. Asks one question — "where do you want to search?" — with three backend lanes (NVIDIA-hosted default, your own URL, or a local deployment), then hands off to /search for the actual query. The journey is the SVG at docs/quickstart-user-journey.svg: Stage 1 (entry) + Stage 2 (backend pick), then /search drives Stages 3-5. Use when: "quickstart", "try usd search", "show me what this does", "getting started", "what is usd search", "tour", "first time".

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

NVIDIA-Omniverse

NVIDIA-Omniverse/usd-search

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

/quickstart — discover USD Search

Your job is the first two stages of docs/quickstart-user-journey.svg:

Open the journey, ask "where do you want to search?"
Pick one of three backend lanes, set USD_SEARCH_API_URL.

Then hand off to /search — it runs Stages 3 (query) → 4 (hybrid search) → 5 (ranked results) per the diagram.

Be terse. No preamble, no per-step narration. The user reads the tool output. Two prompts max — one for the backend, one for the search query (which /search will collect).

Step 1 — Ask "where do you want to search?"

Ask the user once (a single structured question). Three options — these match the three lanes in docs/quickstart-user-journey.svg:

"NVIDIA-hosted" — zero setup, the default lane in the diagram. Set USD_SEARCH_API_URL=https://search.simready.omniverse.nvidia.com. Public NVIDIA dev deployment; answers unauthenticated requests. Proceed to Step 2.
"Your own deployment" — the user already has a USD Search instance. Ask for the name of a shell env var holding the URL (and optionally a token / Basic auth pair). Never accept a pasted secret. Probe curl ${URL}/info/backend/storage; on non-200, ask for a different var or fall back to the NVIDIA-hosted default.
"Local deployment" — one-command setup. Hand off to the deploy-usdsearch skill with "Locally with docker compose" already selected. The deploy skill must run its Local branch, including L1 pre-flight and references/local-runbook.md L2-L6, before returning. When it returns it will have set USD_SEARCH_API_URL for you. Then proceed to Step 2.

Step 2 — Hand off to /search with a seed query

Hand off to the search skill with a suggested seed query. Offer four one-click nouns plus an "Other" escape so the "I don't know what to try" path is one click and a specific request is exact:

"a chair"
"a robot"
"a warehouse"
"a vehicle"
Other — free-text (e.g. "yellow leather armchair")

The /search skill takes it from there: builds the hybrid query, saves five thumbnails to ./search-results/<slug>/, writes a manifest.json, and inspects the JPEGs visually.

Do not auto-pick the query — the single most personal thing this journey does is reflect the user's own words back as a 3D asset.

Step 3 — Print the "now you can…" list

After /search returns, end with a short pointer list (not a question):

/search — refine with a different query, image, or filter
/inspect-asset — deep-dive a single result
/search-in-scene — spatial / scene-graph queries
/deploy-usdsearch — run USD Search yourself

The manifest at ./search-results/<slug>/manifest.json lets these skills pick up where you left off.

Important rules

Two prompts max. Q1 (backend) + Q2 (query, asked by /search).
Never auto-pick the search query.
Check the env before asking for keys. Before prompting for any *_API_KEY or URL token, grep ~/.zshrc ~/.zshenv ~/.zprofile ~/.profile ~/.bashrc ~/.bash_profile ~/.env* for export <NAME>=. If absent, ask the user to export it themselves — never accept a pasted secret.
Indirect credentials. When asking for a URL or token, accept only the name of a shell env var that already holds the value.
Three lanes, not four. Match the SVG: NVIDIA-hosted (default), your own URL, local deployment. No fourth "I have an OpenSearch index" branch — that's an advanced flow that lives in /deploy-usdsearch if it's ever needed.
Hand off, don't reimplement. /search owns the query, the scoring config, the thumbnail fetch, the visual verification. This skill only chooses the backend.

同仓库更多 Skills

同仓库

deploy-usdsearch

NVIDIA-Omniverse/usd-search

Stand up a USD Search deployment the user controls. Branches on local docker-compose vs. helm on Kubernetes. Local branch covers the full quickstart compose stack with GPU/VLM/storage-backend questionnaires. Helm branch covers the usdsearch chart at helm/usdsearch. Use when: "deploy usdsearch", "run the stack myself", "helm install usdsearch", "start the stack", "run usdsearch locally", "host my own".

2026-07-089

explorer-remote-proxy

NVIDIA-Omniverse/usd-search

Run the locally-built Explorer UI against a REMOTE, fully-hosted USD Search instance by proxying every API call through the local gateway. No local deepsearch-api / OpenSearch / SigLIP2 / Asset Graph Service — only the Explorer + gateway run locally, so search, thumbnails, dependency graphs, and downloads all work exactly as on the hosted instance (which already has a populated index + AGS + storage). Use when: you want to test/demo the local Explorer with WORKING dependency graphs and downloads but don't want to (or can't) build a local index + Asset Graph Service. Default target is the public staging instance (api.staging.deepsearch-horde.nvidia.com).

2026-07-089

inspect-asset

NVIDIA-Omniverse/usd-search

Deep-inspect a single 3D asset by URL. Fetches multi-view thumbnails, per-plugin indexing status, USD scene summary (prim counts, geometry, scale, up-axis, default prim), prim hierarchy, forward + inverse dependency graphs, and optionally runs VLM relevance validation against a description. Use when the user asks to inspect / deep-dive / describe / learn about / report on a specific asset, or asks "what's in this USD", "tell me about this asset", "where is this asset used", "what does this asset reference", "is this asset indexed".

2026-07-089

search-in-scene

NVIDIA-Omniverse/usd-search

Spatial and scene-graph queries inside a USD scene via the Asset Graph Service. Answers: what's near a point or prim (radius), what's inside a bounding box, what prims of a given type or property exist in this scene, what does this scene reference, where is this asset used, what's the prim hierarchy. Optionally chains to /search_hybrid with image_similarity_search to find assets similar to a prim's source asset. Use when the user has a scene URL plus a spatial, structural, or dependency question.

2026-07-089

NVIDIA-Omniverse/usd-search

Search USD Search for 3D assets. One skill, three input modes: text query, reference image, or both combined. Hits POST /search_hybrid + GET /images. Iterates on intermediate matches for visual-similarity refinement when the user wants "more like this". Use when: "search assets", "find a 3D model", "find a USD", "find similar", "more like this", "visually similar".

2026-07-089

usd-property-catalog

NVIDIA-Omniverse/usd-search

Discover which USD properties a USD Search deployment's corpus actually carries, and turn that into local config: hits GET /search/stats/usd_properties, then builds a ranked usd_property_catalog.yaml (the grounding source for the built-in LLM query parser) plus ready-to-use search_fields.generated.yaml filter stanzas. Has a P0 "gap audit" mode that reports which target properties are present/absent and pulls sample assets for parsing validation. Use when: "what USD properties can I filter on", "discover usd properties", "build a property catalog", "which metadata is in the index", "audit P0 metadata", "ground the query parser on real properties".

2026-07-089

name	quickstart
license	Apache-2.0
version	3.0.0
description	Get a first-time user to a real 3D asset in seconds. Asks one question — "where do you want to search?" — with three backend lanes (NVIDIA-hosted default, your own URL, or a local deployment), then hands off to /search for the actual query. The journey is the SVG at docs/quickstart-user-journey.svg: Stage 1 (entry) + Stage 2 (backend pick), then /search drives Stages 3-5. Use when: "quickstart", "try usd search", "show me what this does", "getting started", "what is usd search", "tour", "first time".
triggers	["quickstart","try usd search","show me what this does","getting started","what is usd search","tour","first time"]
allowed-tools	AskUserQuestion, Bash, Read, Skill

/quickstart — discover USD Search

Your job is the first two stages of docs/quickstart-user-journey.svg:

Open the journey, ask "where do you want to search?"
Pick one of three backend lanes, set USD_SEARCH_API_URL.

Then hand off to /search — it runs Stages 3 (query) → 4 (hybrid search) → 5 (ranked results) per the diagram.

Be terse. No preamble, no per-step narration. The user reads the tool output. Two prompts max — one for the backend, one for the search query (which /search will collect).

Step 1 — Ask "where do you want to search?"

Ask the user once (a single structured question). Three options — these match the three lanes in docs/quickstart-user-journey.svg:

"NVIDIA-hosted" — zero setup, the default lane in the diagram. Set USD_SEARCH_API_URL=https://search.simready.omniverse.nvidia.com. Public NVIDIA dev deployment; answers unauthenticated requests. Proceed to Step 2.
"Your own deployment" — the user already has a USD Search instance. Ask for the name of a shell env var holding the URL (and optionally a token / Basic auth pair). Never accept a pasted secret. Probe curl ${URL}/info/backend/storage; on non-200, ask for a different var or fall back to the NVIDIA-hosted default.
"Local deployment" — one-command setup. Hand off to the deploy-usdsearch skill with "Locally with docker compose" already selected. The deploy skill must run its Local branch, including L1 pre-flight and references/local-runbook.md L2-L6, before returning. When it returns it will have set USD_SEARCH_API_URL for you. Then proceed to Step 2.

Step 2 — Hand off to /search with a seed query

Hand off to the search skill with a suggested seed query. Offer four one-click nouns plus an "Other" escape so the "I don't know what to try" path is one click and a specific request is exact:

"a chair"
"a robot"
"a warehouse"
"a vehicle"
Other — free-text (e.g. "yellow leather armchair")

The /search skill takes it from there: builds the hybrid query, saves five thumbnails to ./search-results/<slug>/, writes a manifest.json, and inspects the JPEGs visually.

Do not auto-pick the query — the single most personal thing this journey does is reflect the user's own words back as a 3D asset.

Step 3 — Print the "now you can…" list

After /search returns, end with a short pointer list (not a question):

/search — refine with a different query, image, or filter
/inspect-asset — deep-dive a single result
/search-in-scene — spatial / scene-graph queries
/deploy-usdsearch — run USD Search yourself

The manifest at ./search-results/<slug>/manifest.json lets these skills pick up where you left off.

Important rules

Two prompts max. Q1 (backend) + Q2 (query, asked by /search).
Never auto-pick the search query.
Check the env before asking for keys. Before prompting for any *_API_KEY or URL token, grep ~/.zshrc ~/.zshenv ~/.zprofile ~/.profile ~/.bashrc ~/.bash_profile ~/.env* for export <NAME>=. If absent, ask the user to export it themselves — never accept a pasted secret.
Indirect credentials. When asking for a URL or token, accept only the name of a shell env var that already holds the value.
Three lanes, not four. Match the SVG: NVIDIA-hosted (default), your own URL, local deployment. No fourth "I have an OpenSearch index" branch — that's an advanced flow that lives in /deploy-usdsearch if it's ever needed.
Hand off, don't reimplement. /search owns the query, the scoring config, the thumbnail fetch, the visual verification. This skill only chooses the backend.