Research Notes & the Archive
Everything released since 2023 — frameworks, prompt libraries, the curriculum, and the NI-NOTION methodology series. Ideas in progress; read accordingly.
The NI-NOTION Series
The complete methodology, written as it happened — dated entries documenting the system being built and run live.
The Tethered Hyperspace Workflow
The substrate. How to build, classify, structure, execute, evolve, externalize, scaffold, collapse, and cycle structured knowledge domains — with the human as the source of ground truth. The foundation everything else rides on.
Full series available to Insiders.
The Meta-Agent Hyper-Cycling Workflow
The operational layer. The full persistent-agent lifecycle — versioned agent evolution, progressive complexity ramping, deterministic rule encoding — demonstrated live across ten days, running on five reusable tools and one human skill.
Full guide available to Insiders.
Hyperspace Engineering
The mechanical reality of surrogate modeling with language models in latent space — the theory underneath context, agentic, and knowledge engineering.
A language model's latent space is a real geometry, and everything the NERO Method does is an operation on it. Hyperspace Engineering is the name for the mechanics: how meaning is positioned and traversed, what the base model's projection gate makes possible, and how predictable surrogate layers are built on top of it. Five principles carry most of the weight.
Latent space is a geometry
Tokens map to positions; meaning is distance; attention is navigation; generation is a trajectory. This is the model's actual architecture, not a metaphor for it — and it's why structure in the input becomes structure in the output.
Origin shapes the trajectory
The prompt sets the starting position, and the starting position disproportionately determines everything downstream. Prior context creates momentum; stable anchors keep navigation on course. Most of prompt engineering is choosing the origin well.
The gate, then the stack
The vendor's fine-tuned system layer acts as a projection gate that makes interpretation consistent. That consistency is what lets you stack surrogate models on top — each one a predictable layer in the space that makes behavior beneath it more predictable than the raw model.
Specialize; don't generalize
You can specialize the space in any direction you can master. You cannot generalize it without paying somewhere else — context limits and forgetting are the bill. Surrogate modeling is specialization done deliberately, which is exactly what works.
Coherence is load-bearing
When a surrogate's conditions and data are mutually consistent, simulation is accurate. Introduce one inconsistency and incoherence propagates through the projection space and degrades everything downstream. Most failed systems fail here, not on model quality.
Bottom-up, from mastery
Because the space rewards coherence, architecture is built upward from a mastered domain rather than imposed from above — and meta-functionality emerges progressively as the surrogate matures. That is the NERO Method stated at the physics level.
The Hyperspace Codex · 600+ pages · 2024
The complete reference for Hyperspace Engineering — the principles above in full, with the frameworks built on them. Taught live to Insiders as one of the method's two substrates, alongside Band Mechanics.
The Release Archive
Public, dated, cumulative. Selected releases below — the complete library lives in the Insider program.
Frameworks & Models
SELECTED · 2024–2025Earlier research-era frameworks are retired from public presentation; the full 33+ release lineage is preserved in the Insider archive.
Prompt Libraries
16 COLLECTIONS · 2023–2024Curriculum & Documentation
2024 →Open Source, In the Open
27 public repositories, 560+ stars — including the Advanced-GPTs suite at 344★. The public side of the archive lives on GitHub, dated and versioned.
Want the Full Library?
The complete archive — every release, the Codex, the series, and the room where the current work happens — lives in the Insider program.