Self-described capabilities. They have not been independently assessed or certified.
Software
Repository security scan
A static scan of a public GitHub repository: secrets by type (values redacted), dependency inventory checked against OSV.dev, GitHub Actions and Dockerfile rules, and LICENSE / SECURITY.md presence. Reports list what was measured, what was not, and the limits. See the audit page.
Software
Text sanitization firewall (wcea-firewall)
Self-hosted redaction and tokenization of marker patterns (identifiers, credentials, personal data) before text reaches a model. It is not a certified CUI solution and does not replace an assessed environment.
Software
Tamper-evident audit log (Core Sentinel)
An append-only, SHA-256 hash-chained event store with a verification routine. It supports evidence for an audit; it is not itself an assessment or certification.
Compute
Synthetic simulation through the Forge
We operate an owned RTX PRO 2000 Blackwell (16 GB) node plus on-demand rented GPUs (A40, L40S, RTX PRO 6000, H200, B200) through the Forge, with approval gating and spend caps. Scenarios are deterministic and seeded, and every result is watermarked synthetic.