{ "@context": "https://schema.org", "@id": "https://aggr.aymericbeaumet.com/items/simonwillison-net/2026-09-22-llm-typesafe-0-1a0/#webpage", "@type": [ "WebPage", "ArchiveComponent" ], "articleBody": "I built this new plugin for [LLM](https://llm.datasette.io/) to add support for [TypeSafe AI's new Jev model](https://simonwillison.net/2026/Sep/21/jev/). Install it like this:\n\n```\nllm install llm-typesafe\n```\n\nThen set an API key ([get one here](https://console.typesafe.ai/), the waitlist seems to move pretty fast):\n\n```\nllm keys set typesafe\n# Paste key\n```\n\nAnd now you can ask yes/no \"noul\" questions like this:\n\n```\nllm -m jev 'Please refund my last payment.' \\\n -s 'Does this message explicitly request a refund?'\n```\n\nOutput:\n\n```\n{\"type\": \"noul\", \"noul\": 0.99}\n```\n\nOr choice questions like this:\n\n```\ncat message.txt | llm -m jev \\\n -s 'Which team should handle this message? If billing and technical issues both occur, choose billing.' \\\n -o answer_type choice \\\n -o criteria '{\n \"billing\":\"Charges, invoices, payments, or refunds\",\n \"technical\":\"Problems installing or using the product\",\n \"other\":\"Neither category fits\"\n }'\n```\n\nOr scoring questions like this:\n\n```\ncat report.txt | llm -m jev \\\n -s 'How reproducible is the problem described in this report?' \\\n -o answer_type score \\\n -o criteria '[\n \"No reproduction instructions\",\n \"Some instructions, but important steps are missing\",\n \"Complete steps with expected and actual results\"\n ]'\n```\n\nSee [the README](https://github.com/simonw/llm-typesafe/blob/main/README.md) for more details.\n", "authors": [], "captured_at": "2026-09-22T18:47:44.447844335+00:00", "category": "programming", "content_html": "

I built this new plugin for LLM to add support for TypeSafe AI's new Jev model. Install it like this:

\n
llm install llm-typesafe\n
\n

Then set an API key (get one here, the waitlist seems to move pretty fast):

\n
llm keys set typesafe\n# Paste key\n
\n

And now you can ask yes/no "noul" questions like this:

\n
llm -m jev 'Please refund my last payment.' \\\n  -s 'Does this message explicitly request a refund?'\n
\n

Output:

\n
{"type": "noul", "noul": 0.99}\n
\n

Or choice questions like this:

\n
cat message.txt | llm -m jev \\\n  -s 'Which team should handle this message? If billing and technical issues both occur, choose billing.' \\\n  -o answer_type choice \\\n  -o criteria '{\n    "billing":"Charges, invoices, payments, or refunds",\n    "technical":"Problems installing or using the product",\n    "other":"Neither category fits"\n  }'\n
\n

Or scoring questions like this:

\n
cat report.txt | llm -m jev \\\n  -s 'How reproducible is the problem described in this report?' \\\n  -o answer_type score \\\n  -o criteria '[\n    "No reproduction instructions",\n    "Some instructions, but important steps are missing",\n    "Complete steps with expected and actual results"\n  ]'\n
\n

See the README for more details.

\n", "content_markdown": "I built this new plugin for [LLM](https://llm.datasette.io/) to add support for [TypeSafe AI's new Jev model](https://simonwillison.net/2026/Sep/21/jev/). Install it like this:\n\n```\nllm install llm-typesafe\n```\n\nThen set an API key ([get one here](https://console.typesafe.ai/), the waitlist seems to move pretty fast):\n\n```\nllm keys set typesafe\n# Paste key\n```\n\nAnd now you can ask yes/no \"noul\" questions like this:\n\n```\nllm -m jev 'Please refund my last payment.' \\\n -s 'Does this message explicitly request a refund?'\n```\n\nOutput:\n\n```\n{\"type\": \"noul\", \"noul\": 0.99}\n```\n\nOr choice questions like this:\n\n```\ncat message.txt | llm -m jev \\\n -s 'Which team should handle this message? If billing and technical issues both occur, choose billing.' \\\n -o answer_type choice \\\n -o criteria '{\n \"billing\":\"Charges, invoices, payments, or refunds\",\n \"technical\":\"Problems installing or using the product\",\n \"other\":\"Neither category fits\"\n }'\n```\n\nOr scoring questions like this:\n\n```\ncat report.txt | llm -m jev \\\n -s 'How reproducible is the problem described in this report?' \\\n -o answer_type score \\\n -o criteria '[\n \"No reproduction instructions\",\n \"Some instructions, but important steps are missing\",\n \"Complete steps with expected and actual results\"\n ]'\n```\n\nSee [the README](https://github.com/simonw/llm-typesafe/blob/main/README.md) for more details.\n", "dateCreated": "2026-09-22T18:47:44.447844335+00:00", "date_created": "2026-09-22T15:54:16+00:00", "date_modified": "2026-09-22T15:54:16+00:00", "date_published": "2026-09-22T15:54:16+00:00", "description": "Release: llm-typesafe 0.1a0 I built this new plugin for LLM to add support for TypeSafe AI's new Jev model. Install it like this: llm install llm-typesafe Then set an API key (get one here, the waitlist seems to move pretty fast): llm keys…", "external_url": "https://simonwillison.net/2026/Sep/22/llm-typesafe/", "git": { "commit": "ebb0941c5c06a2ab2377342d607535262a3e969e", "path": "items/simonwillison-net/2026/09/2026-09-22-llm-typesafe-0-1a0.md", "permalink": "https://github.com/aymericbeaumet/aggr-instance/blob/ebb0941c5c06a2ab2377342d607535262a3e969e/items/simonwillison-net/2026/09/2026-09-22-llm-typesafe-0-1a0.md" }, "id": "urn:aggr:item:c5612837cb7832b4294001751991ad14efc2a09b", "isBasedOn": { "@id": "https://simonwillison.net/2026/Sep/22/llm-typesafe/" }, "mainEntity": { "@id": "https://simonwillison.net/2026/Sep/22/llm-typesafe/", "@type": "CreativeWork", "archivedAt": { "@id": "https://aggr.aymericbeaumet.com/items/simonwillison-net/2026-09-22-llm-typesafe-0-1a0/#webpage" }, "dateModified": "2026-09-22T15:54:16+00:00", "datePublished": "2026-09-22T15:54:16+00:00", "headline": "llm-typesafe 0.1a0", "timeRequired": "PT1M", "url": "https://simonwillison.net/2026/Sep/22/llm-typesafe/", "wordCount": 173 }, "name": "Archived snapshot: llm-typesafe 0.1a0", "preview": null, "reading_minutes": 1, "replicated_at": null, "source": "simonwillison-net", "source_name": "Simon Willison's Weblog", "summary": "Release: llm-typesafe 0.1a0 I built this new plugin for LLM to add support for TypeSafe AI's new Jev model. Install it like this: llm install llm-typesafe Then set an API key (get one here, the waitlist seems to move pretty fast): llm keys set typesafe # Paste key And now you can ask yes/no \"noul\" questions like this: llm -m jev 'Please refund my last payment.' \\ -s 'Does this message explicitly request a refund?' Output: {\"type\": \"noul\", \"noul\": 0.99} Or choice questions like this: cat message.txt | llm -m jev \\ -s 'Which team should handle this message? If billing and technical issues both occur, choose billing.' \\ -o answer_type choice \\ -o criteria '{ \"billing\":\"Charges, invoices, payments, or refunds\", \"technical\":\"Problems installing or using the product\", \"other\":\"Neither category fits\" }' Or scoring questions like this: cat report.txt | llm -m jev \\ -s 'How reproducible is the problem described in this report?' \\ -o answer_type score \\ -o criteria '[ \"No reproduction instructions\", \"Some instructions, but important steps are missing\", \"Complete steps with expected and actual results\" ]' See the README for more details. Tags: projects, llm, jev", "tags": [ "jev", "llm", "projects" ], "temporalCoverage": "2026-09-22T18:47:44.447844335+00:00", "title": "llm-typesafe 0.1a0", "url": "https://aggr.aymericbeaumet.com/items/simonwillison-net/2026-09-22-llm-typesafe-0-1a0/", "word_count": 173 }