[Integrations](https://spiral.ai/integrations) › [Hermes](https://spiral.ai/integrations/hermes) › [Gemini content classification](https://spiral.ai/integrations/hermes/gemini-content-classification)

# How to Use Gemini content classification in Hermes

Gemini evaluates supplied text, images, audio, video, or PDFs and returns yes/no probabilities, category decisions, or rubric scores. Use it to classify content or grade evidence against explicit rules.

|  |  |
| --- | --- |
| Model | Gemini content classification |
| Available through | Gemini |
| Category | Data |
| Price | Input $0.13 per 1M tokens; no output tokens. |
| Returns | json |
| Runs | In the background |
| Setup | About 2 minutes, no API key |

## Prompts to try with Gemini content classification in Hermes

- Each morning, retrieve new NASA Perseverance announcements and use Gemini content classification on each announcement to choose science finding, rover operations, mission planning, or other. Include the candidate probabilities in the digest.
- Daily, retrieve new pandas-dev/pandas issues and use Gemini content classification on each report to choose documentation, installation, runtime error, or other. Put decisions below 80 percent confidence in a review queue.
- Weekly, retrieve Apple's newly published annual report if one is available and use Gemini content classification to score its risk factors for supply chain relevance from 0 for unrelated to 3 for directly relevant.
- Weekly, retrieve Stripe's webhook documentation and use Gemini content classification to estimate whether it supports guaranteed event delivery order. Report the probability of yes beside the previous evaluation.
- Monthly, retrieve Python's Errors and Exceptions tutorial and use Gemini content classification to score recovery examples from 0 for absent to 3 for demonstrated with runnable code. Compare the returned rubric scores over time.

## How to set up Gemini content classification in Hermes

Pick the way that suits you. Each takes about two minutes.

### Use the Spiral skill

1. **Send Hermes this message**: Your agent reads Spiral's skill and sets itself up.

```
set up https://spiral.ai/SKILL.md
```

2. **Sign in or paste a key**: Approve Spiral in the browser, or give the agent a key from the console.

### Use Spiral's MCP server

1. **Add the server and sign in**: Hermes opens your browser: sign in to Spiral and click Authorize. No API key. When Hermes asks Enable all tools? \[Y/n/select], answer Y: n cancels without saving the server, and select hides the tools you leave out.

```bash
hermes mcp add spiral \
  --url https://api.spiral.ai/mcp \
  --auth oauth \
  --connect-timeout 300
```

2. **Check it**: In a chat that was already open, run `/reload-mcp`.

```bash
hermes mcp list
hermes mcp test spiral
```

### Use the Spiral CLI

1. **Install and sign in**: Installs the Spiral CLI, signs you in and adds its skill to your coding agents.

```bash
npx -y spiral-cli@latest init
```

### Call the Spiral API

1. **Create an API key**: In the console, under API keys.

2. **Run Gemini content classification**: One request; the result comes back by polling or a webhook.

```bash
curl https://api.spiral.ai/v1/runs \
  -H "Authorization: Bearer $SPIRAL_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{
    "provider": "gemini",
    "model": "gemini-decision-1",
    "input": {
      "input": "<input>",
      "questions": {}
    }
  }'
```

Then paste one of the prompts above: Hermes finds Gemini content classification on Spiral, shows the price and runs it.

Setup guide in the docs: https://spiral.ai/docs/quickstarts/hermes

## Gemini content classification tools and prices

- [Gemini Decision 1](https://spiral.ai/tools/gemini/gemini-decision-1) (Gemini): Input $0.13 per 1M tokens; no output tokens.

## What can you do with Gemini content classification in Hermes?

Produce a daily queue of public announcements and issue reports classified against fixed editorial rules. Evaluate each newly retrieved item with Gemini and use returned probabilities to decide what enters the review queue.

### Route public bug reports

Classify supplied issue text into documentation, installation, runtime error, or other, then use candidate probabilities to prioritize review.

### Sort research announcements

Assign supplied NASA announcements to mission updates, research findings, launch operations, or other for an editorial queue.

### Grade documentation coverage

Score a supplied technical guide against defined levels for prerequisites, examples, error handling, and troubleshooting.

### Label public product photos

Evaluate supplied catalog images for visible packaging, product views, and background characteristics using separate yes/no questions.

### Check claims against supplied evidence

Pair a draft product claim with its technical documentation and estimate whether the supplied evidence supports the claim.

### Triage public company disclosures

Classify supplied filing excerpts by operational, regulatory, cybersecurity, or supply chain risk and score relevance to a defined research brief.

## Overview

- [Google Gemini](https://spiral.ai/providers/gemini)
- [All Hermes guides](https://spiral.ai/integrations/hermes)

## Gemini content classification in other agents

- [Amp](https://spiral.ai/integrations/amp/gemini-content-classification)
- [ChatGPT](https://spiral.ai/integrations/chatgpt/gemini-content-classification)
- [Claude](https://spiral.ai/integrations/claude-desktop/gemini-content-classification)
- [Claude Code](https://spiral.ai/integrations/claude-code/gemini-content-classification)
- [Cline](https://spiral.ai/integrations/cline/gemini-content-classification)
- [Codex](https://spiral.ai/integrations/codex/gemini-content-classification)
- [Cursor](https://spiral.ai/integrations/cursor/gemini-content-classification)
- [Devin](https://spiral.ai/integrations/devin/gemini-content-classification)
- [dots](https://spiral.ai/integrations/dots/gemini-content-classification)
- [Factory Droid](https://spiral.ai/integrations/factory-droid/gemini-content-classification)
- [Gemini](https://spiral.ai/integrations/gemini/gemini-content-classification)
- [Gemini CLI](https://spiral.ai/integrations/gemini-cli/gemini-content-classification)
- [Google Antigravity](https://spiral.ai/integrations/antigravity/gemini-content-classification)
- [goose](https://spiral.ai/integrations/goose/gemini-content-classification)
- [Grok](https://spiral.ai/integrations/grok/gemini-content-classification)
- [Grok Build](https://spiral.ai/integrations/grok-build/gemini-content-classification)
- [Instinct](https://spiral.ai/integrations/instinct/gemini-content-classification)
- [Manus](https://spiral.ai/integrations/manus/gemini-content-classification)
- [Muse](https://spiral.ai/integrations/muse/gemini-content-classification)
- [Muse Code](https://spiral.ai/integrations/muse-code/gemini-content-classification)
- [n8n](https://spiral.ai/integrations/n8n/gemini-content-classification)
- [Notion](https://spiral.ai/integrations/notion/gemini-content-classification)
- [OpenClaw](https://spiral.ai/integrations/openclaw/gemini-content-classification)
- [opencode](https://spiral.ai/integrations/opencode/gemini-content-classification)
- [Perplexity](https://spiral.ai/integrations/perplexity/gemini-content-classification)
- [Raycast](https://spiral.ai/integrations/raycast/gemini-content-classification)
- [VS Code](https://spiral.ai/integrations/vscode/gemini-content-classification)
- [Windsurf](https://spiral.ai/integrations/windsurf/gemini-content-classification)
- [Zapier](https://spiral.ai/integrations/zapier/gemini-content-classification)

## More Hermes guides

- [Firecrawl research paper metadata and passages](https://spiral.ai/integrations/hermes/firecrawl-research-paper-metadata-and-passages)
- [GitHub Repository](https://spiral.ai/integrations/hermes/github-repository)
- [Firecrawl related research papers](https://spiral.ai/integrations/hermes/firecrawl-related-research-papers)
- [GitHub contribution activity](https://spiral.ai/integrations/hermes/github-contribution-activity)
- [Firecrawl paper search](https://spiral.ai/integrations/hermes/firecrawl-paper-search)
- [GitHub trending developer list](https://spiral.ai/integrations/hermes/github-trending-developer-list)

## Questions about Gemini content classification in Hermes

**Can Hermes use Gemini content classification?**

Yes. Add Spiral to Hermes as an MCP server, and Hermes can find and run Gemini content classification like any built-in tool.

**How much does Gemini content classification cost in Hermes?**

Input $0.13 per 1M tokens; no output tokens. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your Hermes plan is separate.

**Do I need a Gemini account to use Gemini content classification in Hermes?**

No. Spiral runs Gemini content classification on its own Gemini account. You sign in to Spiral once and never paste a key.

**How long does it take to set up Gemini content classification in Hermes?**

About two minutes: add the server and sign in, then check it.

**Does Hermes wait while Gemini content classification runs?**

Gemini content classification runs in the background. Hermes starts the run, checks on it and gives you the json when it's ready.

**Where does the content being evaluated come from?**

You supply the content Gemini evaluates. Google documents support for text, images, audio, video, and PDFs in one shared input.

**How much content can Gemini evaluate in one request?**

Google documents a context limit of 256K tokens and up to 1,000 independent questions against the shared input.

**Does a yes/no result include uncertainty?**

Gemini returns an estimated probability of yes between 0.01 and 0.99, plus confidence. The API leaves the decision cutoff to your application.

**Is a rubric score always a whole number?**

The returned score is a probability-weighted average, so it can fall between rubric levels. The response also includes the legend, probabilities, and confidence.

**Can a request return no classifications?**

Yes. Google's reference states that blocked input returns no answers.

