Skip to main content

jevgram

jevgram is a small command-line tool that checks if a document is AI-generated. Similar to pangram but a lot cheaper. You need API key from TYPESAFE_API_KEY (super easy, couple of clicks)

Supported inputs

  • Plain text and source-like text files, such as .txt, .md, .tex, and similar files
  • PDF files
  • DOCX files

Minimal example:

$ export TYPESAFE_API_KEY=apikey_...
$ pipx install jevgram
$ jevgram README.md # this document
File: README.md
Type: text
Characters extracted: 1957
Characters sent to JEV: 1957
Model: jev-1.13.0
Prediction: AI
Probability AI: 87.00%
Probability human: 13.00%
Confidence: 74.00%

PDF extraction uses pypdf. DOCX extraction is handled directly from the Word XML contained in the .docx file.

Detailed instructions

Install

With pipx (MacOS, Linux):

pipx install jevgram

With uvx:

uvx jevgram --help

API key

jevgram uses TypeSafe JEV API. Set one of these environment variables before running it:

export TYPESAFE_API_KEY="your-key"
# or
export JEV_API_KEY="your-key"

You can also pass a key directly with --api-key, but environment variables are safer for shell history and scripts.

Usage

jevgram document.txt
jevgram paper.pdf
jevgram manuscript.docx --json

By default, long files are sampled down to 32,000 characters using the beginning, middle, and end of the extracted text. To send the whole extracted document, use:

jevgram document.txt --max-chars 0

To inspect extraction without calling JEV:

jevgram document.pdf --extract-only

To inspect the request body without calling JEV:

jevgram document.txt --dump-request

Output

Human-readable output includes the predicted class, AI probability, human probability, confidence when provided by JEV, and whether the input was truncated. --json produces machine-readable output for scripts.

Cost

Once you get a TYPESAFE account and API key, you get a credit for $5 free usage. jevgram is intended for personal use, and you can check a lot of documents for $5. Once you use up the free credit, jevgram will fail and TYPESAFE will probably ask for credit card number (I have not reached that point yet)

Notes

Authorship detection is probabilistic.

This document is mostly AI :-)

Metadata

Release files for jevgram 0.1.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for jevgram 0.1.2
File Size Uploaded Provenance
jevgram-0.1.2.tar.gz 8.8 kB Without provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for jevgram 0.1.2
File Interpreter ABI Platform
jevgram-0.1.2-py3-none-any.whl Python 3 none any Details

Total release size: 18.9 kB

Release files / jevgram-0.1.2.tar.gz

Download URL jevgram-0.1.2.tar.gz
Size 8.8 kB
Tags Source
SHA-256 checksum
How to use checksums
150d2f345c83af08732cd5e015b5c4e01ad35235195ffc2d6254fa3da2c7f8f0
BLAKE2b-256 checksum
How to use checksums
9ce4498d590a1384ddb97b42030b4c9dfc08b17f6e1f276f7e21863d2b710c8a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.8.4

Release files / jevgram-0.1.2-py3-none-any.whl

Download URL jevgram-0.1.2-py3-none-any.whl
Size 10.2 kB
Tags Python 3
SHA-256 checksum
How to use checksums
48d148f9b5e8bfbb4e5a36bee34ca0bbd4436bcc9143ef82e08ebfce48e5b81b
BLAKE2b-256 checksum
How to use checksums
3a8360d0f21ec4758a984c4ab0982a729d1c6e75aca02beb702fa6961db3b64c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.8.4

Security for jevgram 0.1.2

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page