1. Made by
“The code is open source under the MIT licence at”
The server is part of the classifier.dev service itself; the repo owner runs the classifier.dev domain, its terms point to this repo, and the registry name dev.classifier sits on that domain.
classifier.dev/terms2. Keys
“claude mcp add --transport http classifier https://classifier.dev/mcp”classifier.dev/mcp-setup
3. Runs
Endpoint: https://classifier.dev/mcp. Checked: yes: an unauthenticated GET to https://classifier.dev/mcp on 2026-11-10 answered 405, as documented for this POST-only server.
classifier.dev/mcp4. Commercial use
“The short version: the service is free within the published limits, it is offered as it is, you are responsible for what you send and for what you do with the answers, and it can change or stop.”
Free within the published limits and offered as is; you are responsible for what you send and for using the answers.
classifier.dev/termsChecked against its repository at one commit, on Nov 10, 2026. Tap a verdict to see its source.How we checkedReport an error
What classifier.dev does
classifier.dev lets your AI assistant sort a pile of texts, like support tickets, reviews or search results, into labels you name, such as bug, feature or praise. Each answer comes with a confidence number, so you read only the unsure ones. It works without an account or key, but by default the texts you send are kept to train its models unless you switch that off.
What people use it for
Sort a backlog of support tickets
Your assistant puts each ticket into billing, bug or question and gives a confidence for each, so the sure ones are routed and you read the rest.
Tag articles with every topic that fits
An article can be about pricing and security at once. The assistant gets every label that applies to each text, with a score per label.
See what share of feedback is bugs
Get only the totals, such as 412 bugs, 230 praise and 58 unsure, without pulling a thousand answers into the chat.
Find the items worth a human look
After a big sort, ask for only the texts the model was unsure about, each with its second guess, and read just those.
Route each ticket by team and urgency
One request labels every ticket by several things at once, for example which team, how urgent and what kind, each with its own confidence.
Filter search results before reading them
Your assistant found forty links; it labels them relevant or not and opens only the six that matter, which saves time and tokens.
classifier.dev popularity
Tools it gives your agent (6)
What it sends and its license
the texts, labels, instructions and results of every classification
To classifier.dev (private Cloudflare storage), kept for model training and operator review; every classifying tool call without share_data false. Opt out: pass "share_data": false in the tool arguments of every call.
the texts and labels
To TypeSafe (directly or through Vercel AI Gateway); for smart re-asks, a reasoning model via OpenRouter; every classification. Opt out: none for the hosted service; run the MIT code yourself.
the caller's IP address (for abuse checks on free use)
To Spur's Context API; free requests, on a cache miss. Opt out: use a funded workspace key.
The code is open source under the MIT license. Using the hosted service is free within its limits and offered as is. The answers are yours, but texts you send are kept for training unless you opt out.
Free within the published limits and offered as is; you are responsible for what you send and for using the answers.
“The short version: the service is free within the published limits, it is offered as it is, you are responsible for what you send and for what you do with the answers, and it can change or stop.” source
Send only client texts you may share; without share_data false they are kept for training.
“You must have permission to contribute any content you share for training.” source
The server code is MIT, so you may copy and run it yourself with the license notice kept.
“The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.” source
The answers are yours to use however you like, including as training data for your own model; your inputs are kept for classifier.dev's own training unless you pass share_data false.
“The answers are yours to keep and to use however you like.” source
Is it alive
1.0.0 in the registry, updated Sep 19, 2026
1.0.0 in the MCP Registry (published 2026-09-19); code commit b0e3b44 of 2026-10-19
Alternatives
classifier.dev This page
flair
Pick it if the texts must not leave your own servers
scikit-learn
Pick it if you already have labelled examples and want to train your own model
Pick it if you only need to read classifier.dev's docs, limits or prices
Other servers from the same repository
classifier.dev docs
Get started
Tell your AI assistant
“Connect https://classifier.dev/mcp, then sort the texts in feedback.txt into bug, feature request and praise with classify_texts, always passing share_data: false, and show me only the answers under 0.7 confidence.”
Connect it
claude mcp add --transport http classifier https://classifier.dev/mcp{"mcpServers": {"classifier": {"url": "https://classifier.dev/mcp"}}}{
"mcpServers": {
"classifier-dev": {
"url": "https://classifier.dev/mcp"
}
}
}Mistakes to avoid
An omitted setting counts as sharing, so the texts are kept for training.
Answers re-asked by the smart tier come back with null confidence and scores.
Every call returns one of your labels; the label names are all the model has to go on.
Public smart batches above 200 texts are refused.
The server's own instructions say it only pays off when reading the inputs is the expensive part.
Questions
Do I need an account or a key?
No. The public server works without a key within per-address limits; a workspace key only adds paid capacity and usage records.
Are my texts kept?
Yes by default, for training and operator review. Pass share_data false on each call to opt out; signing in or paying does not opt you out.
Where do my texts go?
To classifier.dev, which forwards them to its model providers, TypeSafe and, for the smart tier, a reasoning model through OpenRouter.
How many texts can one call take?
Up to 1,000 texts and 2 to 100 labels per call.
Is there a separate server for its documentation?
Yes, at classifier.dev/mcp/docs; it reads and searches the docs and classifies nothing.