Seed

Last updated 12 Sep 2026

Seed scans authorized local folders for checksum-valid recovery phrases and recognized private-key formats. It reports candidates; validation does not prove ownership, value, or relevance.

Scan scope and defaults

Choose one source folder and a local output folder. The default thread count is half the available CPU parallelism, with a minimum of one. The default maximum file size is 50 MiB, the uncommon-word threshold is 4, and the result limit is unlimited. Exclusion patterns can omit matching paths.

Text and wallet categories are on by default. Their extensions are:

txt log json env csv md xml html htm cfg ini yaml yml toml conf config bak backup
note notes sql dump dat key keys wallet seed mnemonic recovery secret

Document scanning is off by default. When enabled, it reads PDF, DOCX, XLSX, and XLSM; documents larger than 8 MiB are skipped by the extractor. Turn every category off and the scan will refuse to start instead of silently scanning another category.

Archives and images

ZIP scanning is off by default. When enabled, Seed reads up to 10,000 entries per archive and nested ZIPs to a maximum depth of two. Each extracted member is still subject to the configured file-size limit.

Image OCR is off by default and requires an active sign-in. It accepts JPG, JPEG, PNG, GIF, BMP, and WebP images up to 1 MiB. Images are sent to the Krantz API for hosted OCR, with at most four requests in flight and up to three attempts. All non-image scanning stays local.

What is detected

  • checksum-valid English BIP39 recovery phrases
  • raw secp256k1 hexadecimal keys and WIF keys
  • extended private keys with xprv, yprv, or zprv prefixes
  • Solana private keys in base58 or JSON-array form
  • Nostr private keys with the nsec prefix

Identical phrases and keys are published once per run. The result view includes the source path, line, word count, and uncommon-word score when available.

Output

Each scan creates a timestamped directory under the configured Seed output folder:

  • seeds.txt and keys.txt contain deduplicated raw findings.
  • csv/seeds.csv adds word count, score, source file, and line.
  • csv/keys.csv adds key kind, source file, and line.
  • csv/files.csv records each file’s status, reason, size, and result counts.
  • summary.txt and csv/summary.csv record totals, skips, failures, bytes, and duration.

Unreadable, oversized, excluded, binary, empty-extraction, disabled-image, and result-limit skips are counted separately. Review every candidate in context and never send a phrase or private key to support.