# AI use cases for Classifier Dev (mrmps/classifier-dev)

Repo: [mrmps/classifier-dev](https://github.com/mrmps/classifier-dev) | Use cases: 3 | Page: https://dedupedai.com/use-cases/mrmps-classifier-dev/

A zero-shot text classifier served over plain HTTP, with a CLI and an MCP server, that needs no API key or account.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Sort a batch of texts into labels in one call: https://dedupedai.com/use-cases/mrmps-classifier-dev/sort-texts-into-labels/
- Re-check low-confidence labels with a stronger model: https://dedupedai.com/use-cases/mrmps-classifier-dev/escalate-uncertain-labels/
- Send a structured bug report when a label looks wrong: https://dedupedai.com/use-cases/mrmps-classifier-dev/report-classifier-problems/

## How to sort a batch of texts into labels in one call with Classifier Dev

Repo: [mrmps/classifier-dev](https://github.com/mrmps/classifier-dev) | Difficulty: Intermediate | Time: 15 minutes | Basis: From the README

Label short texts with plain HTTP, no key or signup, up to a thousand per request.

**Who it is for:** A developer or ops person with a list of messages, tickets or comments to triage and no time to set up a model.

**The problem:** Hand-sorting or hand-writing rules for a pile of short texts takes hours and breaks as wording changes.

**The outcome:** Each text gets a label and a calibrated confidence, returned as plain text that a script or agent can read.

### Steps

1. Run `curl https://classifier.dev/spam,not+spam/Win+a+free+iPhone` to see the path form return a label.
2. Run `curl "https://classifier.dev/?labels=spam,not+spam&text=Win+a+free+iPhone"` to see the same call in query form, which is easier to build in code.
3. Run `curl classifier.dev` to read the plain text docs for the batch form, since the batch syntax is not shown in the README excerpt.
4. Send your texts in batches of up to a thousand per request using the batch form from the docs, and collect each label in order. (inferred, check the README)
5. Write the input text and returned label side by side so you can spot-check the results.

### Prompt to paste into your agent

```text
Using the classifier.dev API from the classifier-dev repo, label each text in <input file> as bug, feature or praise by calling https://classifier.dev over plain HTTP with no key. Write the results to <output file> as CSV with the original text and the label. Verify that the number of output rows equals the number of input rows.
```

### Caveat

The README excerpt shows the single-text form and states the thousand-text limit, but not the exact batch syntax, so confirm it at classifier.dev before scripting.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/mrmps-classifier-dev/sort-texts-into-labels/

## How to re-check low-confidence labels with a stronger model with Classifier Dev

Repo: [mrmps/classifier-dev](https://github.com/mrmps/classifier-dev) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README

Use the smart tier so answers under 0.7 confidence get a second opinion.

**Who it is for:** A data or ops person whose batch of labels includes some ambiguous items that a fast answer may get wrong.

**The problem:** Fast single-label answers are less reliable when the model is unsure, and it is hard to tell which ones to recheck by hand.

**The outcome:** Uncertain single-label answers are replaced by a reasoning model's answer and marked escalated, while confident answers stay as they were.

### Steps

1. Read the tier notes in the docs with `curl classifier.dev` to confirm how the tier option is set on a request.
2. Add the smart tier to the request for single-label classification. (inferred, check the README)
3. Check each result for `escalated: true` to find the answers that were replaced.
4. Compare the escalated answers against a small hand-checked sample before trusting the whole batch.

### Prompt to paste into your agent

```text
Using the classifier-dev repo at classifier.dev, classify each item in <input file> with the smart tier for single-label work. Output a table of item, label, confidence and whether it was escalated. Verify that every escalated row has a label that was returned by the service.
```

### Caveat

The README says smart re-asks single-label answers below 0.7 but does not show the parameter name, and multi-label requests ignore the tier, so verify the request format first.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/mrmps-classifier-dev/escalate-uncertain-labels/

## How to send a structured bug report when a label looks wrong with Classifier Dev

Repo: [mrmps/classifier-dev](https://github.com/mrmps/classifier-dev) | Difficulty: Advanced | Time: 30 minutes | Basis: From the README

Let an agent file a feedback.now report when classification breaks or misbehaves.

**Who it is for:** An AI agent that gets wrong or failing classifications and needs to tell the maintainers without a human writing the report.

**The problem:** A wrong or failing result goes unreported, so the same problem keeps happening for every caller.

**The outcome:** A structured report reaches the maintainers, and the agent gets a receipt showing whether it landed and how good it was.

### Steps

1. Run `curl https://classifier.dev/.well-known/agent-feedback.json` to see what the host accepts. (inferred, check the README)
2. Run `curl https://classifier.dev/api/v1/policy` to read the categories, severities and limits. (inferred, check the README)
3. Send a full structured report to `POST /api/v1/feedback` with the category, severity and the input and output that show the problem.
4. Keep the report id from the response and check `GET /api/v1/receipts/{id}` to confirm it was accepted. (inferred, check the README)

### Prompt to paste into your agent

```text
Using the feedback.now endpoints at https://classifier.dev described in the classifier-dev repo, file one report for <describe the wrong label> with the input text and the returned label as evidence. Output the report id and the receipt status. Verify that the receipt shows the report was accepted or marked as a duplicate.
```

### Caveat

Accepted reports are emailed to the maintainers and kept for 90 days, so include only the input and output needed to show the problem and no secrets.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/mrmps-classifier-dev/report-classifier-problems/