How is this different from exporting rows to a Python script that calls an LLM?

You ask where the data already lives instead of pulling tables out and writing answers back. Jev picks from your criteria set so the result is typed by construction, and one request carries many questions together.

Will a bad criteria literal really fail before scanning a huge table?

Yes. Empty maps, duplicates, single-level rubrics, more than 255 options, or non-constant criteria fail when the query is planned, and queries without a secret also fail at plan time.

Why did the same Jev call in WHERE and SELECT bill twice?

DuckDB evaluates a function once per place it appears, so the same call in two clauses bills twice per row. Identical requests are cached for the life of the process, and rows in a chunk go out sixteen at a time.

What happens on rate limits or dropped connections mid-query?

Rate limits, server errors, and dropped connections retry with backoff while other errors fail at once. jev_usage reports requests, cache hits, and tokens, and jev_on_error set to null keeps a bad row from failing the query.

How do I supply the API key?

With CREATE SECRET (TYPE jev, API_KEY 'sk-...'); ENDPOINT and MODEL are optional, and with no secret queries fail when planned rather than part-way through.

What SQL types do the functions return?

jev_choice returns an ENUM of your options, jev_score a DOUBLE on the scale, jev_noul a DOUBLE for the probability of yes, and jev_ask a STRUCT with one field per question; choice and score also carry a name_confidence field.

Can I keep one failing row from killing the query?

Yes. Set jev_on_error = 'null' to lose the row instead of the query; otherwise rate limits, server errors, and dropped connections retry with backoff while any other error fails at once.

CommunityJev Integrations & Adapters

Duckdb Jev — colliber

DuckDB extension that asks a Jev question about every row in SQL and returns a typed ENUM, DOUBLE, or STRUCT column instead of generated text.

See highlights

Open github.com

Stars: 29
Language: C++
License: MIT
Published by: Community
Author: colliber
Last verified: Oct 2026

colliber/duckdb-jev

Organization repository on GitHub

View on GitHub

DuckDB extension: typed Jev answers as real SQL types

Forks: 1
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

Ask a question about every row of a table, in SQL, and get a real SQL type back. A DuckDB extension over Jev, TypeSafe's model for typed answers instead of text. The data you want to ask about already sits in a table or a Parquet file, and SQL is the query language everyone has. So ask the question where the data is, instead of pulling it out, wrapping an API in a script and writing the answer back.

Quoted from the project's README. Read the full README.

Tagscommunity jevbest catalog-import c++

Works forJev for Classification

Used inJev for Software & DevTools

Highlights

jev_choice, jev_score, jev_noul, and jev_ask map criteria literals to ENUM, DOUBLE, and STRUCT columns with confidence fields.
Criteria double as the allowed answers and column type, must be constant, and fail at plan time on empty, duplicate, or oversized sets.
Sends sixteen rows per request with process-lifetime caching, backoff retries, and jev_usage() token accounting.

Watch out

Needs a TypeSafe key via CREATE SECRET and costs about one request per row; SET jev_on_error to 'null' drops failing rows instead of failing queries. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

Other listings of this project

Forks and copies of this project in the directory. This entry is the canonical one.

duckdb-jev

Native C++ DuckDB extension that runs Noul, Choice, Score, and mixed judgments directly from SQL with batching, concurrent HTTP, and chunked streaming and no Python UDFs.

Jev Integrations & Adapters#community#jevbest#catalog-import

Community

prasanthj/duckdb-jevC++

More Jev resources like this

duckdb-jev

Native C++ DuckDB extension that runs Noul, Choice, Score, and mixed judgments directly from SQL with batching, concurrent HTTP, and chunked streaming and no Python UDFs.

Jev Integrations & Adapters#community#jevbest#catalog-import

Community

prasanthj/duckdb-jevC++

system-one

Provider-neutral TypeScript runtime for System One decisions: write choice, noul, and score code once, then run it on TypeSafe Jev, local Reflex, any compatible /v1/systemone endpoint, or a custom provider.

Jev Integrations & Adapters#community#jevbest#catalog-import

Community

iamaamir/system-oneTypeScript

tocket

File-first project notebook where many agents share context in .context/ files while Jev only judges the next step and records it as JSON.

Jev Integrations & Adapters#community#jevbest#catalog-import

Community

pedrocivita/tocketTypeScript

jev-tree

Typed Python planner runtime that turns batched Jev action distributions into executable probability trees and graphs for finite decision tasks.

Jev Integrations & Adapters#community#jevbest#catalog-import

Community

Chuf-H/jev-treePython

Astra

Self-hosted agent runtime that adds an optional Jev judgment layer to memory filtering, lesson dismissal, request classification, and skill selection, with inspectable context and execution.

Jev Integrations & Adapters#community#jevbest#catalog-import

Community

matrixorigin/AstraRust

loki

Agent runtime with a terminal UI and a messaging gateway that keeps memory, self-built skills, session search, scheduling, and subagents across Telegram, Discord, Slack, WhatsApp, Signal, and CLI.

Jev Integrations & Adapters#community#jevbest#catalog-import

Community

wundercorp/lokiPython

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai