Machine-native computing · Open-source research

Designing data, storage, and policy boundaries for verifiable computing.

Project Meno is an open-source research effort exploring machine-native computing: structuring state, storage, and execution boundaries around deterministic verification and bounded context rather than around unstructured prose and ambient authority.

Research status & transparency note: Project Meno is active experimental research preparing for staged open-source release, not production-ready commercial software. Reference prototypes have been written and cross-checked with AI models under James Everson’s architectural direction and logic spot-checking; systematic human line-by-line review and independent third-party security review are planned as components advance through release gates.

Thesis

Machine-native computing and the Meno reference implementation.

Most contemporary software and AI agent tooling inherits conventions built solely for manual human editing: sprawling markdown files, unvalidated configuration prose, full-file re-reads, and permissive shell execution. When automated tools and language models operate over those surfaces, token costs climb, context degrades, and security checks rely on brittle regular expressions.

Machine-native computing treats deterministic structure as the canonical representation and treats human-readable documents as rendered views. Project Meno is one open-source reference approach to that discipline ⁠— designed so other researchers and engineers can test its ideas, adopt what holds up under measurement, and build better implementations freely.

Research areas

Four interconnected technical layers.

  • Zero-code structured representations (.mnd, .mnv): Stack-depth-zero line grammars and non-secret vocabulary tables written and sliced through deterministic tools, then projected into clean human-readable documentation and code views.
  • Exhaustively decidable policy gates (.mni): Finite boolean and enumerated gates (bounded to at most 16 inputs) whose entire 2^k state space is verified against totality and mutation checks, enforcing strict separation between data and executable code.
  • Content-addressed storage and relational indexing (MNFS): Immutable content-addressed object stores, deduplicated volume snapshots, and content-blind relational graph indexes that resolve topology without scanning opaque or sensitive file contents.
  • Bounded AI agent execution and continual learning: Context-metered slicing, polite command rerouting versus hard security denies, hardware-key gates on irreversible actions, and tiered rule promotion with multi-session cooling-off periods.

Read the architectural overview →

Open-source model

Public research, independent of proprietary lock-in.

Project Meno is not a proprietary software product. Its components are being cleaned, audited for license and provenance clarity, and prepared for public open-source release in waves ⁠— paired with an explicit evidence ledger distinguishing externally benchmarked results from synthetic prototypes and analytical designs.

Developers, researchers, and organizations will be free to use Meno under an open-source license with or without its maintainers. For teams seeking direct workshops, readiness assessments, or custom architecture support, independent professional services are offered through JREonline LLC.

Release roadmap & verification standards · Adoption & consulting