How to Use Nemotron 3.5 Content Safety in Claude Code
Use Nemotron 3.5 Content Safety to classify user requests and AI replies as safe or unsafe. It supports text-and-image moderation, custom safety policies, and optional reasoning traces for reviewing decisions.
Prompts to try with Nemotron 3.5 Content Safety in Claude Code
Copy an example into Claude Code, or ask in your own words.
Use Nemotron 3.5 Content Safety to evaluate the moderation fixtures in this project. Compare each verdict with its expected label and report false positives and false negatives separately.
Use Nemotron 3.5 Content Safety to check the saved user requests and draft assistant replies in this project. Preserve separate verdicts for requests and responses, and list disagreements with our expected results.
Use Nemotron 3.5 Content Safety to evaluate our customer support examples under the project's policy. Check threats, personal information exposure, and social engineering while allowing harmless complaints.
Use Nemotron 3.5 Content Safety to test the custom disclosure policy in this repository. Compare examples requesting confidential material with examples asking about published documentation.
Use Nemotron 3.5 Content Safety to compare the current moderation policy with the proposed revision using our existing labeled examples. Report which classifications changed and whether each change matches the expected label.
How to set up Nemotron 3.5 Content Safety in Claude Code
Pick the way that suits you. Each takes about two minutes.
Use Spiral's MCP server
- Add the hosted server
This adds Spiral to the current project. Add --scope user to use it in every project.
claude mcp add --transport http spiral https://api.spiral.ai/mcp
- Sign in
Start claude, type /mcp, choose spiral and Authenticate. Sign in to Spiral in the browser and click Authorize. No API key.
Use the Spiral skill
- Send Claude Code this message
Your agent reads Spiral's skill and sets itself up.
set up https://spiral.ai/SKILL.md
- Sign in or paste a key
Approve Spiral in the browser, or give the agent a key from the console.
Use the Spiral CLI
- Install and sign in
Installs the Spiral CLI, signs you in and adds its skill to your coding agents.
npx -y spiral-cli@latest init
Call the Spiral API
- Create an API key
In the console, under API keys.
- Run Nemotron 3.5 Content Safety
One request; the result comes back by polling or a webhook.
curl https://api.spiral.ai/v1/runs \ -H "Authorization: Bearer $SPIRAL_API_KEY" \ -H "Content-Type: application/json" \ -H "Idempotency-Key: $(uuidgen)" \ -d '{ "provider": "openrouter", "model": "nvidia/nemotron-3.5-content-safety", "input": { "messages": [ { "role": "system" } ] } }'
Then paste one of the prompts above: Claude Code finds Nemotron 3.5 Content Safety on Spiral, shows the price and runs it.
Nemotron 3.5 Content Safety providers and prices
Spiral runs Nemotron 3.5 Content Safety through OpenRouter. Open a card for today's price, with every fee included.
Up to $0.001076 per run at the default inputs. $0.26 per 1M input tokens, $0.26 per 1M output tokens.
Settings you can ask for
Ask Claude Code for any of these in your prompt.
| Setting | Allowed values | Default | Description |
|---|---|---|---|
| Seed | a whole number | ||
| Min p | 0 to 1 | ||
| Top k | a whole number from 0 | ||
| Top p | 0 to 1 | ||
| Max tokens | 1 to 117964 | The most output tokens to generate. | |
| Temperature | 0 to 2 | ||
| Presence penalty | -2 to 2 | ||
| Frequency penalty | -2 to 2 | ||
| Repetition penalty | 0 to 2 |
Every API field of NVIDIA: Nemotron 3.5 Content Safety is on its tool page: open its card above.
What can you do with Nemotron 3.5 Content Safety in Claude Code?
Build moderation checks and safety regression reports around your project's assistant. Use Nemotron 3.5 Content Safety to classify requests and replies, then compare its decisions with your labeled examples.
Screen incoming requests
Classify user requests against safety rules before passing them to an assistant.
Review draft AI replies
Assess the user request and proposed reply together, keeping their safety verdicts separate.
Triage customer support messages
Apply rules for threats, personal information exposure, and social engineering while explicitly allowing harmless frustration.
Check enterprise assistant disclosures
Use a custom policy to flag requests for confidential information while allowing discussion of published documentation.
Run safety regression checks
Compare classifications with human-labeled examples after changing an assistant or moderation policy.
Evaluate multilingual moderation
Classify equivalent moderation examples across your target languages and compare the results with reviewer judgments.
Review custom policy decisions
Use optional reasoning traces to examine disputed classifications and refine ambiguous rules.
More on Nemotron 3.5 Content Safety and Claude Code
Overview
Nemotron 3.5 Content Safety in other agents
Questions about Nemotron 3.5 Content Safety in Claude Code
Can Claude Code use Nemotron 3.5 Content Safety?
Yes. Add Spiral to Claude Code as an MCP server, and Claude Code can find and run Nemotron 3.5 Content Safety like any built-in tool.
How much does Nemotron 3.5 Content Safety cost in Claude Code?
Up to $0.001076 per run at the default inputs. $0.26 per 1M input tokens, $0.26 per 1M output tokens. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your Claude Code plan is separate.
Do I need an OpenRouter account to use Nemotron 3.5 Content Safety in Claude Code?
No. Spiral runs Nemotron 3.5 Content Safety on its own OpenRouter account. You sign in to Spiral once and never paste a key.
How long does it take to set up Nemotron 3.5 Content Safety in Claude Code?
About two minutes: add the hosted server, then sign in.
Does Claude Code wait while Nemotron 3.5 Content Safety runs?
Nemotron 3.5 Content Safety runs in the background. Claude Code starts the run, checks on it and gives you the text, json when it's ready.
What should I use Nemotron 3.5 Content Safety for?
Use it to classify user requests and AI responses against safety rules. It also fits moderation reviews and safety regression checks.
What does Nemotron 3.5 Content Safety return?
It returns text containing a safety label for the user input and a separate label for an AI response when one is supplied. Optional reasoning traces help you review how it applied a policy.
How much text or image content can it assess?
NVIDIA documents up to 128K tokens of context and at most one image per classification. Its vision encoder resizes images to 896 by 896 pixels.
Does Nemotron 3.5 Content Safety support multilingual moderation?
NVIDIA describes coverage across 12 languages. Evaluate your actual languages and moderation examples before relying on its classifications.
Can its safety verdicts be wrong?
Yes. NVIDIA identifies both false positives and false negatives as known risks, so test allowed content alongside prohibited examples and review disagreements.
Start using Nemotron 3.5 Content Safety in Claude Code
One Spiral account for every tool your agent needs. No subscription; you pay per run.
Paste this into Claude Code:
set up https://spiral.ai/SKILL.md


