- Category
- Developer Tools
- Rank
- No. 1369Tools index
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Data Processing
- Interfaces
- CLI
- GitHub
- 61 stars
- Latest release
- v0.4.0
- Date
About
grev is a set of open-source, dual-licensed (Apache-2.0/MIT) Unix filter tools that answer typed natural-language questions about each line of input, using TypeSafe's Jev models to return calibrated probabilities instead of matching regex patterns. This lets commands behave like grep, sort, or cut but filter or label logs, commit messages, or text by meaning, at a cost of fractions of a cent for thousands of lines.
At a glance
grev filters lines by asking a decision model a question about each input. It is worth trying when you want to find the same event described in several ways and a text pattern keeps missing examples.
Good fit
Command-line users screening synthetic logs or other text they can share with a model provider.
Consider something else when
Skip it when a known literal pattern covers the job or your inputs cannot leave your environment.
Start here
Create a few synthetic log lines that describe a timeout in different words and include a similar line about a successful retry. Check whether the filter returns the lines you expected before using it in a pipeline.
Cost and requirements
Apache-2.0 project. It uses TypeSafe Jev for scoring, so live calls require provider access and incur usage according to that account.
Limits to know
The provider receives input text for scoring. Meaning-based matches can omit relevant lines or include near matches, so review the result before using it to make an automated decision.
Related work
- JevMan (context)
Sources checked 2026-11-05. Source-based assessment.
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.