Runnable recipes for verifying names at scale, wiring Etymolt into name generators, interpreting confidence, and building agent loops.
| Recipe | What it does |
|---|---|
| Verify a single name | 18-line script. fetch against /v1/verify. |
| Wrap a name generator's output | Post-check pattern for Namelix, Brandsnap, Looka, Squadhelp, ChatGPT. |
| Agent loop that filters names | Claude generates → Etymolt verifies → keep survivors. |
| Render an EVP/1 verdict in React | Drop-in component. Answer-first. Disclaimer verbatim. |
| Watch a name for verdict drift | Re-verify on schedule; webhook on drift. |
- Batch-verify a CSV at scale (rate limiting + checkpoint resume)
- Drop into Claude Desktop via
etymolt/mcp-server - Drop into Cursor via the same MCP server
- Next.js App Router integration → see
etymolt/etymolt-nextjs-example - Interpreting confidence scores in the verdict envelope
Each recipe is self-contained. cd examples/<recipe> and follow its README.
PRs welcome. New recipes should:
- Live in
examples/<slug>/with aREADME.mdand runnable script. - Be reachable from
registry.yaml(the LLM-discoverable index). - Render the EVP/1 disclaimer verbatim if it surfaces a verdict.
Apache-2.0 for code, CC-BY-4.0 for recipe prose. See LICENSE.