| Version | Supported |
|---|---|
| v0.1.x | Yes |
| < 0.1 | No |
If you believe you have found a security issue in the SemanticMap spec itself, in the example documents, in CI tooling, or in any release artifact:
- Do not open a public GitHub issue.
- Email
security@unboxapi.prowith a clear description and any proof-of-concept. v0.1.0 accepts plaintext email; a PGP release key (docs/pgp.asc) ships in v0.1.1 once Founder + CTO have minted and cross-signed it. Until then, please mark sensitive details as such in the subject line so triage can route appropriately. - We will acknowledge within 2 business days and aim to provide an assessment within 5 business days.
- Coordinated disclosure window: 90 days from acknowledgment, extended by mutual agreement if a fix requires more time.
CTO is the first responder. CEO is informed of any High/Critical report within 24 hours of triage.
The full threat-model memo lives on the source issue and covers:
- Schema-injection by malicious SemanticMap producers (T1).
- Parser abuse against consumer YAML/JSON parsers (T2).
- Supply-chain integrity of this repository's published artifacts (T3).
- Accidental information disclosure (T4).
- Abuse of the open repository (T5).
A SemanticMap document is data, not code. Consumers must treat untrusted SemanticMap documents the same way they treat any other untrusted input.
- Branch protection on
main: required PR review, required CI status checks, no direct pushes, no force-push, linear history. - Required signed commits (Sigstore
gitsignor GPG). Release tags signed. - Sigstore artifact attestation on every release.
- CycloneDX SBOM published as a release asset.
- CODEOWNERS requires CTO review on every PR.
- Dependabot, GitHub secret scanning, and GitHub Advanced Security code scanning enabled.
gitleaksruns on every PR and on the full commit range at release.