User repository on GitHub
use jev in dbt
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Integrations & Adapters
Dbt Jev
A dbt package that evaluates SQL values with Jev through TypeSafe's hosted API or OpenRouter: macros expose Choice as a nullable label, Noul as a match probability and Score as a numeric rating, on DuckDB and ClickHouse.
- Stars
- 3
- Language
- Python
- License
- MIT
- Reading time
- 2 min
- Published by
- Community
- Author
- smithclay
- Last verified
- Oct 2026
- Platforms
- DuckDBOpenRouter
- Capabilities
- Open sourceRequires API key
Highlights
- Public macros classify, match_probability, score and decisions work unchanged on DuckDB and ClickHouse.
- match_probability sends two text-cast expressions as structured left and right state and returns a number from 0 to 1.
- score uses ordered levels and returns an expected value from 0 through levels minus 1, so a three-level rubric can produce fractional values such as 1.75.
- decisions batches several typed questions about the same input into one Jev request; decision_label and decision_value pull each answer back out.
- Macros only generate SQL, so inference starts when the database executes it; tested with dbt Core 1.11.12, DuckDB 1.5.5 and ClickHouse Server 26.8.9.10.
Quickstart
sql
select
ticket_id,
{{ dbt_jev.score(
'ticket_context',
levels=['No urgency', 'Needs attention', 'Urgent'],
instructions='Rate the urgency of this support request'
) }} as urgency_score
from {{ ref('stg_support__tickets') }}Watch out
Grounded in the truncated repository README and GitHub metadata reviewed on 2026-10-15; the README was cut off part-way through the tested-support table, and no pricing beyond the linked provider pages is documented. All version numbers come from the project's own compatibility table.
Questions about this entry
Short answers drawn from the linked project or page.
A dbt package whose macros evaluate SQL values with Jev through TypeSafe's hosted API or OpenRouter. It exposes Jev Choice as a nullable label, Noul as a nullable match probability and Score as a nullable numeric rating.
The same public macros work on DuckDB and ClickHouse. The README's tested matrix includes dbt Core 1.11.12, dbt-duckdb 1.11.0 with DuckDB 1.5.5, and dbt-clickhouse 1.10.2 with ClickHouse Server 26.8.9.10 on Python 3.12.12.
Use dbt_jev.classify with a column and named choices, for example expected, unexpected and unknown for tool-call failures. It returns a nullable label you can materialise as a table column.
It sends the two text-cast expressions as structured left and right state with instructions and criteria, then returns a number from 0 to 1 for whether the records represent the same customer, which suits candidate-pair dedup after ordinary SQL has generated pairs.
dbt_jev.decisions takes several typed questions about the same input and returns a JSON object mapping each question id to its scalar answer in a single Jev request; decision_label extracts a Choice label and decision_value extracts a Noul probability or Score.
The macros only generate SQL. Inference starts when the database executes the SQL, never while dbt parses or compiles it, so the README recommends materialising decisions as tables so downstream reads use stored answers.
A dbt project plus credentials for one of two paths: TypeSafe AI's hosted API or OpenRouter. The package itself is MIT-licensed, and generating the SQL needs no key at all because inference happens later, at execution time.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
Hacker News: A MySQL plugin that filters rows by asking Jev whether a column matches a sentence.
maayanlevy-hn2 pointsA MySQL plugin that filters rows by meaning (built on TypeSafe Jev)Read the thread on Hacker NewsX: jev(): Postgres WHERE clauses in plain language
More Jev resources like this
Five self-contained Jev demos for data and analytics engineers: a LinkedIn cringe meter, semantic SQL in DuckDB, a SQL VISUALIZE intent, and dbt semantic tests locally and on Databricks, each labelled live or simulated.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
datagobes/jev-demosPython
datagobes/jev-demosPythonA Python library that runs LLM and Jev calls as Polars expressions: parallel inference across providers, Pydantic-validated outputs as struct columns, and a native TypeSafe System One layer for per-row decision questions.
Jev Integrations & Adapters#community#catalog-import#logicrw
Community
pnthn-ai/polar_llamaPython
pnthn-ai/polar_llamaPythonCustom ComfyUI nodes that use Jev for text interpretation and judgments — select candidates, evaluate conditions, score text and extract numbers — plus a node for picking local SKILL.md files.
Jev Integrations & Adapters#community#catalog-import#logicrw
Community
hndrr/comfyui-jevPython
hndrr/comfyui-jevPythonLoadable SQLite C extension and Python package that exposes Jev Noul, Choice, and Score judgments as SQL scalar functions and a jev_rows virtual table which batches up to 40 rows into one shared state and joins typed results back to the source by rowid.
Jev Integrations & Adapters#community#sqlite#sql
Community
mgaitan/sqlite-jevC
mgaitan/sqlite-jevCLakeSail built five Jev functions into Sail's SQL, so Spark SQL and DataFrame queries can ask typed questions about every row, with calls overlapping in the Rust engine behind one in-flight limit per worker.
Jev Integrations & AdaptersArticle#community#sql#rust
Community
A MySQL plugin that filters rows with a natural-language condition. AILIKE asks Jev whether a column value matches, and returns 1, 0, or NULL.
Jev Integrations & Adapters#community#sql#database
Community
maayanlevy/mysql-ailikeC++
maayanlevy/mysql-ailikeC++From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.