A public framework for engineering AI under control.

What this site is

This site is a living knowledge atlas and public reference model for engineering and governing AI-enabled software, models, agents, identities, evidence, and operating authority as one system.

• Sourced concepts with maturity, provenance, failure modes, and implementation signals
• Eight lifecycle phases and four cumulative risk pathways
• Independent verifier and evidence-production models
• A target enterprise AI control-plane architecture
• Sixty-four sourced knowledge terms, implementation patterns, use-case blueprints, and eight source-informed essays
• Four field-curriculum paths across twenty-six finite stages, explicitly separated from certification and authority
• Enterprise and home-lab deployment paths connected by one shared identity, control, verifier, evidence, and authority spine
• A browser-local use-case workbench with portable plans, evidence manifests, and control backlogs
• An Agent System Passport and Assurance Case Studio that compile declared system identity, controls, independent verifier contracts, evidence obligations, control gates, and named human authority into portable draft artifacts
• A browser-local Agent Estate Workspace that compiles portable passports into an as-of-date reference inventory of ownership, standing, risk, verifier coverage, recertification, and unresolved governance attention
• A browser-local Policy Pack Compiler that projects an exact governed Plan and Agent System Passport into deterministic, non-executable CI, runtime-interception, verifier-adapter, and evidence-operations references
• A deterministic Control Plane Lab for inspecting simulated policy, verifier, evidence, and human-disposition paths

What this site is not

The framework is not evidence that every depicted capability is currently activated. It is not a certification, legal conclusion, client control mapping, or substitute for use-case-specific risk and compliance work.

• No fabricated client, partner, or regulator claims
• No implication that a reference architecture proves implementation
• No unsupported compliance, certification, or deployment claims

Why the about route exists

This route gives human readers, crawlers, and AI systems a stable, canonical explanation of the framework, its scope, and its limits. It also separates the target operating model from claims about present activation.

The operating rule

Agents build. Verifiers challenge. Humans authorize. The control plane enforces. In this framework, an agent can execute work, but it cannot accept enterprise risk or approve its own output.

Editorial contributions

The Atlas is curated through an editorial and technical review process, not an open comment stream. Public intake is not open yet. Invited contributors use the private repository workflow and must provide provenance, implementation implications, limitations, disclosure, and publication rights.

• No confidential client information, PHI, credentials, or private vulnerabilities
• Perspectives remain distinct from WHPS policy and implemented capability
• Accepted work retains attribution, source provenance, review date, and version history