shard-core v0.2.0
shard-core 0.2.0
First stable open-source release of shard-core.
Highlights
- Authenticated threshold recovery using ChaCha20-Poly1305 and Shamir sharing
of a random data-encryption key. - Resilient recovery across bounded threshold combinations, including safe
handling of redundant corrupted or unrelated shares. - Explicit-output, atomic, symlink-refusing secret file operations.
- Fordefi recovery-phrase split/combine and direct SHEN encrypt/decrypt
workflows without assuming BIP-39. - Optional SLIP-39 support for material explicitly confirmed as BIP-39.
- Hash-locked, network-disabled, rootless-Podman offline bundle build.
- AI-agent operating guidance in
AGENTS.md, with secrets kept outside the
agent boundary.
Compatibility
- Existing SHRD v1 and v2 artifacts remain readable.
- Existing SHEN v1 and v2 artifacts remain readable.
- No SHRD v1/v2 or SHEN v1/v2 wire-layout change was made for this release.
- Python 3.9 through 3.14 is supported.
Verified source
- Commit:
0d8da202d655f6d53cd97eb833d91b391ed6737e - Git tree:
438aafc4d3a5f08ff79148d05e4aef4e93ab5a14 - Controlled offline candidate smoke:
PASS - Candidate install,
pip check, import smoke, and CLI version:PASS
Install
For a normal source install, use the immutable tag:
git clone --branch v0.2.0 --depth 1 \
https://github.com/Protocol-Wealth/shard-core.git
cd shard-core
python3 -m pip install .For an air-gapped install, download the offline bundle archive and the
top-level checksum file, verify the archive hash, extract it, then follow the
included VERIFY.md and run install-offline.sh.
Release assets
- Python wheel and source distribution.
- Linux x86_64 offline bundle with its own internal
SHA256SUMS. - Top-level
shard-core-0.2.0-SHA256SUMS.txt. - Bundle metadata and provenance.
- Sanitized, non-secret Stage 6 smoke evidence.
Security status
This project has not received an independent cryptographic security audit.
Passing tests and a successful offline build smoke test are release evidence,
not a security audit or a guarantee against loss. Review SECURITY.md,
THREAT_MODEL.md, and CEREMONY.md before using real secret material.
PyPI publication is not part of this GitHub release. The repository includes a
separate least-privilege Trusted Publisher workflow for an optional future
PyPI publication.