NVIDIA AI Blueprint for Video Search and Summarization (VSS) is a reference architecture for applications that use AI to search, analyze, question, and summarize live or recorded video. It is intended for building video analytics agents with vision-language models, language models, retrieval-augmented generation, and NVIDIA NIM microservices. The catalogue entries provide coding-agent rules, skills, commands, agents, instructions, and settings associated with the project.
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.
npx skills add NVIDIA-AI-Blueprints/video-search-and-summarization --skill benchmark-video-searchgit clone --depth 1 https://github.com/NVIDIA-AI-Blueprints/video-search-and-summarizationWrote 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.
[](https://agentmods.dev/skills/nvidia-ai-blueprints/video-search-and-summarization/benchmark-video-search)<a href="https://agentmods.dev/skills/nvidia-ai-blueprints/video-search-and-summarization/benchmark-video-search"><img src="https://agentmods.dev/badge/skills/nvidia-ai-blueprints/video-search-and-summarization/benchmark-video-search/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.
<a href="https://agentmods.dev/skills/nvidia-ai-blueprints/video-search-and-summarization/benchmark-video-search"><img src="https://agentmods.dev/badge/skills/nvidia-ai-blueprints/video-search-and-summarization/benchmark-video-search.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00063 | $0.04152 |
| Opus 5.5 | $0.00025 | $0.01661 |
| Sonnet 5.5 | $0.00013 | $0.00830 |
| Haiku 4.5 | $0.00006 | $0.00415 |
Grade A, and why
benchmark-video-search scanned grade A with 1 finding 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 8d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
| Search profile agent reachable | `curl -sf --connect-timeout 5 --max-time 10 ${ENDPOINT}/health` returns 200 | How it starts
The opening of the file, as written. The whole thing — 302 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Instructions
Follow the routing table, then the numbered steps. Execute each step in order on
a first run; for a repeat run against an already-indexed deployment go straight
to Step 5. Reference material is in references/, the runner is in
scripts/.
Purpose
Answers two questions about a search profile deployment: does retrieval find
the right footage, and where does a query's time go. It drives the vss CLI —
the same path the product uses, since the agent adapter shells out to
vss search run <mode> --raw for every search — so the numbers describe what
ships rather than a REST endpoint that is being retired.
When not to use this skill
- For a single video-search question, use
vss-search-archive. - For an end-to-end benchmark of the OpenClaw chat route, this runner does not measure that path; do not present CLI timings as chat timings.
- For a historical benchmark of the retired agent REST search endpoint, use
the external legacy
run_eval.pynoted inreferences/troubleshooting.md.
Routing
| Situation | Action |
|---|---|
| User asks to benchmark the OpenClaw chat route | This runner does not measure it; explain the gap instead of presenting CLI timings as chat timings |
| User asks to benchmark the retired agent REST search endpoint | This runner does not measure it; for historical comparison, use the external legacy run_eval.py noted in references/troubleshooting.md |
| No search profile deployed in this session | Use vss-build-vision-ai to deploy the stock search profile with its in-stack agent REST API and in-stack LLM retained. Explicitly name both requirements in the build request so it skips the harness question (which would remove the agent and possibly the LLM); do not select a NemoClaw-only or CLI-only harness. Record the reachable agent endpoint and unified origin, then return here. Check agent /health and LLM /v1/models, run Step 3 to inspect indices, and complete Step 4 ingestion and its index probe before scoring |
| User did not give an endpoint | Ask for it. Do not guess, and do not default to localhost |
Endpoint uses localhost or 127.0.0.1 |
This is valid when the runner is on the deployment host. Verify the actual VST clip URL is reachable from RT-VLM before trusting critic-filtered metrics |
| User did not give a dataset | Ask which dataset and where its --data-dir is. Do not invent one |
Elasticsearch has no mdx-* indices |
Ingestion has not completed. Run Step 4; do not report zero scores as a quality result |
| User asks to reuse what is already ingested | Add --skip-ingest. It cannot be combined with --clear or --only-dataset |
| User asks to analyse an existing result file | Skip to Step 6; read the JSON, run nothing |
| Every query returns 0 hits | Stop. Diagnose with Step 3 before reporting anything — this is nearly always missing indices, not poor retrieval |
Critic-filtered metrics are all NA |
The critic never ran. Check the clip-URL prerequisite below before concluding anything about verification quality |
What ships with it
19 files 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.
- .gitignore 266 B
- evals/evals.json 2.3 KB
- references/dataset-format.md 4.3 KB
- references/flags.md 3.6 KB
- references/reading-results.md 9.0 KB
- references/troubleshooting.md 6.3 KB
- scripts/flows/__init__.py 6.8 KB runs code
- scripts/flows/base.py 11 KB runs code
- scripts/flows/dataset.py 12 KB runs code
- scripts/flows/decompose.py 8.3 KB runs code
- scripts/flows/ingest.py 21 KB runs code
- scripts/flows/metrics.py 9.3 KB runs code
- scripts/flows/normalize.py 5.9 KB runs code
- scripts/flows/query.py 15 KB runs code
- scripts/flows/readiness.py 6.4 KB runs code
- scripts/flows/routing.py 8.7 KB runs code
- scripts/run_eval_flows.py 77 KB runs code
- scripts/tests/test_search_eval_flows.py 80 KB runs code
- skill-card.md 1.0 KB
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.
- 8d ago First seen · 302 lines · 63 tokens per session scan A 50d8cf3ec649
benchmark-video-search is a skill published in the GitHub repository NVIDIA-AI-Blueprints/video-search-and-summarization (1,932 stars, last pushed today), licensed Apache-2.0. It adds 63 tokens to every session and 4,152 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-09.
Other skills, from other repositories
agent-platform-rag-engine-management
Manage and query Agent Platform RAG Engine Corpora and retrieve grounded contexts using the Google GenAI SDK. Use when listing RAG corpora or files, inspecting a corpus, retrieving contexts, or generating content grounded in a RAG corpus. Do not use for standard database queries (use SQL/Spanner skills), Google…
graphify
Use for any question about a codebase, its architecture, file relationships, or project content — especially when graphify-out/ exists, where the question should be treated as a graphify query first. Turns any input (code, docs, papers, images, videos) into a persistent knowledge graph with god nodes, community…
google-cloud-solution-rag-enterprise-search-gke-sqldb
Discovers requirements, and generates architectural, design, and deployment guidance for a retrieval-augmented generation (RAG)-capable enterprise search system in Google Cloud. Use when users need a vector-enabled SQL database as the store and index for the embedding vectors, an open model and open-source inferencing…
beautiful-article
A tool for turning source material—such as a web page, PDF, DOCX, Markdown file, text, screenshot, or pasted content—into a single offline HTML article.
gpt-image-2
A skill for generating or editing images with GPT Image 2 across local, host-provided, or advisory setups.
web-design-engineer
Build or redesign polished browser-rendered visual artifacts with HTML/CSS/JavaScript/React: pages, dashboards, prototypes, slide decks, animations, UI mockups, and data visualizations. Use for visual front-end creation, design-system exploration, design critique, or explicit browser acceptance / QA of a web artifact.…