wickra-feature-store is analysis software: it builds data-driven feature
matrices from market data and places no orders. It holds no secret material and
opens no network connections — it reads only the candle data the caller hands it.
The attack surface is therefore narrow — principally the parsing of untrusted
FeatureSpec / universe data as it crosses the C ABI and WASM boundary. See
THREAT_MODEL.md for the asset inventory and trust boundaries.
Security fixes land on main and ship in the next release. 0.2.1 is the
first published release; no earlier version exists to support.
| Version | Supported |
|---|---|
| 0.2.1 (latest) | ✅ |
Please do not open a public issue, pull request or discussion for security problems. Report privately through either channel:
- GitHub → the repository's Security tab → Report a vulnerability (private advisory), or
- email support@wickra.org.
Include a description, affected version/commit, reproduction steps and impact.
We aim to acknowledge within a few days, agree a disclosure timeline, and credit reporters who wish to be named once a fix ships.
In scope: memory-safety or panic-across-FFI flaws in the C ABI hub and its
buffer protocol, denial-of-service through a hostile FeatureSpec or dataset (for
example unbounded allocation while parsing), and any input that makes a binding
return a corrupted or non-deterministic matrix. Out of scope: incorrect indicator
mathematics (a functional bug, not a vulnerability) and advisories in third-party
crates that are already tracked and triaged.
This repository ships a machine-readable VEX record in
osv-scanner.toml, kept in lock-step with the cargo-deny
advisory ignore list in deny.toml. Any advisory assessed as not
affecting wickra-feature-store is documented there with a reason, so downstream
scanners see an explicit, auditable justification rather than an unexplained
suppression.