There are currently 90+ repositories under github.com/issdandavis. That sounds like chaos. It is not — it is a single architecture spread across a constellation of small services, each of which exists because the core SCBE thesis required a concrete artifact to prove it. Everything in the ecosystem follows one pattern: bounded AI in governed loops. This article is the map.
The monorepo. The 14-layer pipeline, the test suites (321+ files, 6066 tests), the Sacred Tongue assembler and VM, the cyber suite modules, the training harness. If a component is too small to warrant its own repo, it lives here. Published under github.com/issdandavis/SCBE-AETHERMOORE.
The Poincaré Hyperbolic Distance Metric model: 21 dimensions broken into 6D hyperbolic position, 6D phase vector, 3D flux, and 6D audit trail. This is the embedding every SCBE component uses to turn data into a cost-space point. Trained via QLoRA on the auto-generated SFT pairs from Polly's auto-conversation loop.
A standalone library of hyperbolic geometry primitives — Poincaré ball operations, Möbius addition, geodesic interpolation, distance formulas. Extracted out of SCBE so anyone can build hyperbolic models without pulling in the whole governance framework.
Bijective tokenization plus context-aware encryption using the Six Sacred Tongues. Geoseal is how every SCBE artifact is signed and how cross-service messages are sealed so that the tongue weight of a message is cryptographically bound to its content.
The lore bible for the Aethermoore narrative universe. It looks like fiction at first glance, but under the hood it is a structured knowledge graph used to generate training data, test inputs, and demo scenarios for the governance framework. The "fiction" is where the vocabulary comes from.
A governance-aware browser runtime. Every tab is a sandbox; every navigation passes through the 14-layer pipeline; every extension has to pay hyperbolic cost to interact with the DOM. Think of it as "what if Chrome had an H(d, R) threshold on fetch?"
A peer-to-peer substrate for distributing SCBE agents across machines. Nodes discover each other, exchange trust-tube parameters, and relay audit trails. Used by the HYDRA swarm and the pope bot.
The oddest and most charming repo in the constellation. The repo is the agent: GitHub Actions workflows are the runtime, issues are the input queue, commits are the side effects. It has no server anywhere. It wakes up when a cron fires, does its job, and goes back to sleep. This is the purest form of the "bounded AI in governed loops" pattern — the loop is literally the CI scheduler.
A fork of a multi-channel personal AI assistant, adapted to route every tool call through the SCBE pipeline and to use the HYDRA swarm for any task that crosses a security boundary. See the HYDRA article for how the six heads split the work.
Every repo in this ecosystem implements the same shape: a small AI component bounded by a hyperbolic envelope, running inside a governed loop that can prove, at every step, that its actions stay within the envelope. The repos are not a pile of side projects — they are the same design instantiated against different substrates.
One-builder ecosystems usually collapse into a single giant monorepo, because managing 90 repos is a nightmare. SCBE keeps them separate for a specific reason: each repo is a proof artifact. If the same pattern can work as a browser runtime, as a CI-driven agent, as a crypto library, and as a distributed P2P network, then the pattern is load-bearing, not cosmetic. Splitting into many repos also means each can be licensed, audited, and consumed independently — which matters when you are pitching the framework to entities that do not want to swallow the whole universe at once.
If you want the short path through the ecosystem, start with SCBE-AETHERMOORE, then read the 14-layer pipeline article, then branch to whichever satellite matches your interest. Commercial services built on top of the ecosystem are on the services page.