Skip to content
@etymolt

etymolt

The fact-check layer for LLM-generated names.

Etymolt

The fact-check layer for LLM-generated names.

Trademark · domain · cultural · sound · pronunciation — signed verdicts across five canonical axes.

Product · Methodology · Protocol · Security


Try it

curl -X POST https://api.etymolt.com/v1/verify \
  -H "content-type: application/json" \
  -d '{"name":"Stratagem"}'

The response is a signed EVP/1 verdict. Free tier, no signup.

Use it from your stack

MCP — Claude, Cursor, ChatGPT, Windsurf @etymolt/mcp-server
OpenAPI — generate any client etymolt/openapi
Recipes — examples for every surface etymolt/cookbook
Brand & design — as installable agent skill etymolt/design-skills

Read the protocol

EVP/1 is the open (CC-BY-4.0) wire format for signed brand-name clearance verdicts. Five canonical axes, Ed25519 signatures over JCS-canonicalized payloads, verbatim disclaimer, temporal-validity fields. In public comment until 2026-09-10.

etymolt/evp-spec

What a verdict is

A clearance signal compiled from records of record. Not legal advice. Not a recommendation. Not a name generator. We report what the registries show; counsel decides what to do about it.

Naming, attested.

hello@etymolt.com · @etymolt

Pinned Loading

  1. cookbook cookbook Public

    Etymolt cookbook — runnable recipes for verifying names at scale, wiring into name generators, interpreting confidence, building agent loops.

  2. design-skills design-skills Public

    Etymolt design tokens, brand voice, and UI primitives — installable as an agent skill via npx skills add etymolt/design-skills.

  3. etymolt-nextjs-example etymolt-nextjs-example Public

    Drop-in Next.js App Router example for the Etymolt API. Deploy to Vercel in one click. Uses @etymolt/sdk.

    TypeScript

  4. evp-spec evp-spec Public

    Etymolt Verdict Protocol — open specification for signed brand-name clearance verdicts. CC-BY-4.0. In public comment until 2026-09-10.

    Python 1

  5. llm-hallucination-benchmark llm-hallucination-benchmark Public

    Public ground-truth corpus measuring how often GPT-5, Claude 4.7, Gemini 3, and Llama 4 fabricate USPTO citations for brand names. Updated quarterly.

    Python

  6. mcp-server mcp-server Public

    Official MCP server for Etymolt — the fact-check layer for LLM-generated names. Drop-in tool for Claude Desktop, Cursor, Windsurf, ChatGPT, and any MCP-compatible host.

    TypeScript

Repositories

Showing 10 of 14 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…