In plain English
This page is part of the technical reference. It keeps the expert detail but starts with a plain-language summary for first-time readers.
- Why this matters: AI risk can come from the whole arrangement, not one obvious model.
- What to look for: data, memory, routes, adapters, tools, evaluators, updates, and rollback paths.
- Technical version below: the expert terminology remains available and is linked through the glossary.
Apex Credibility Finalization v1.21.3
This implementation record closes the Apex Threat credibility/source-expansion prompt as a package release. It preserves the existing Apex concept while making the public claim boundary easier to verify through visible source cards, real-instance proof limits, implementation controls, and exact evidence labels.
Scope
- Bumped the active site, theme, package, discovery, test, and
.uairelease surfaces to v1.21.3. - Finalized the Apex Threat real-world analogue teaser so LLM03, LLM06, and LLM08 are all visible from the main Apex page.
- Preserved the five new Apex credibility pages:
/apex-threat/real-instances/,/apex-threat/source-map/,/apex-threat/evidence-levels/,/apex-threat/implementation-controls/, and/apex-threat/what-is-not-proven/. - Preserved the source data files and reusable Apex include components introduced by the source-expansion pass.
- Added a finalization validator for package versioning, source counts, teaser completeness, Apex navigation, external-link safety attributes, and
.uaifinal-report continuity. - Refreshed generated content indexes, route inventory, sitemap, Atom feed, AI advisory files, manifests, integrity hashes, and UAIX memory metadata.
Boundary
This record is not a live deployment check, certification, peer review, vulnerability disclosure, or threat-intelligence advisory. The full Apex Threat ecology remains presented as a system-design synthesis from documented component risks, not as a confirmed named malware family, CVE, or single real-world incident.
Verification targets
Use these visible routes to verify the bounded evidence model:
/apex-threat/source-map//apex-threat/evidence-levels//apex-threat/real-instances//apex-threat/implementation-controls//apex-threat/what-is-not-proven/