Forge: a zero‑dependency secret scanner

Описание и применение

Forge – это одностраничный Python‑CLI без сторонних зависимостей, реализующий сканирование секретов, расчёт энтропии, оценку достоверности и устойчивое хранилище, уменьшая размер установки до 0 Б и добавляя рейтинг найденных секретов.

Задача

Existing secret scanners like detect‑secrets and truffleHog depend on many third‑party packages, inflate install size, and lack confidence scoring for findings.

Результат

Forge achieves a 0‑dependency, 78 KB CLI that matches or exceeds the functionality of detect‑secrets while reducing install footprint from ~4 MB to 0 bytes and adding per‑finding confidence scores.

Рабочий процесс

• 1. Implement regex rules and entropy calculation using re and collections.Counter.
• 2. Build a naive‑Bayes confidence scorer with hand‑tuned weights and sigmoid output.
• 3. Create an append‑only log‑structured storage layer with zlib.crc32 checksums and atomic compaction.
• 4. Integrate scanning, confidence scoring, and storage into a ten‑subcommand CLI using argparse.
• 5. Package the tool as a single .pyz file and verify zero runtime dependencies via pip install in a fresh venv.

Доказательства

Forge is a single-file Python CLI with zero third‑party dependencies…

Источник

The scanner is the story… both halves are stdlib problems…

Источник

Forge delivers a single‑file CLI (78 KB) with zero third‑party dependencies…

Источник