Modus Primus: Engineering Specification for AI Architecture (PAHA Companion)

Modus Primus is the engineering specification companion to the Portable
Agent Harness Architecture (PAHA) pattern. PAHA establishes the architectural
pattern for sovereignty-bounded enterprise AI deployments; Modus Primus
establishes the engineering practice that produces and maintains
PAHA-conformant AI architectures.

The specification addresses engineering concerns that general systems
engineering practices do not fully cover: substrate volatility, behavioral
non-determinism, federated governance across security enclaves, and runtime
assurance properties. It provides architectural principles (the means and
mechanisms selection relation, the five-M decomposition), a Work Breakdown
Structure for a Modus Primus instance, a five-tier federation pattern
(Modus Primus through Modus Quintus) with inheritance, delegation, and
escalation contracts, four V&V instruments specialized for AI architecture,
seven domain-specific technical review gates, and a per-enclave tech
baseline pattern that separates architectural specification from realization.

The specification is integration-ready for two enterprise contexts:
enterprises with an existing systems engineering management plan layer the
AI-specific specializations on top of their existing practice; enterprises
without one treat the specification as the engineering specification
directly. Enterprise-specific decisions are flagged with [ENTERPRISE:]
placeholders for editorial completion.

Version 1.1 incorporates adoption refinements over the initial publication
release, including an external standards mapping in Section 1.1, a
PAHA-to-Modus correspondence table in Section 1.3, a First Actions by
Audience subsection (Section 1.6), an Out of Scope subsection (Section 1.7),
and a worked means election example in Section 6.3.

The specification cites Portable Agent Harness Architecture (Longmire 2026,
Revision 2.2; doi:10.5281/zenodo.20112631) as its architectural framework.