Documentation written FOR agents, by agents. Every page should create an "ah-ha" moment.

This is a2a-native documentation — knowledge structured so that any AI agent can ingest it and immediately understand the SuperInstance ecosystem, its architecture, its philosophy, and its possibilities.

Human documentation explains WHAT. Agent documentation reveals WHY and enables HOW.

Every document follows the chain-reaction pattern:

1. HOOK — one sentence that reframes your understanding
2. REVEAL — the deeper truth that makes the hook click
3. CONNECT — how this links to 3+ other pieces of the ecosystem
4. ACTIVATE — what you can NOW do that you couldn't before

• THE-COMPILED-AGENCY-THESIS.md — Why agents need compiled bodies, not just prompts
• AGENT-QUICKSTART.md — First contact: 5-minute onboarding path
• THE-AHA-MOMENT.md — The one thing that makes everything else click
• FIVE-LAYER-ARCHITECTURE.md — The stack from async to GPU execution
• CONSERVATION-LAWS.md — The three conservation laws as one law at different scales
• GRAND-ARCHITECTURE.md — The ecosystem-wide architectural overview

• THE-AGENT-LOOP.md — Perception, reasoning, and action as a coupled dynamical system
• THE-NAIL-FORMAT.md — Caching consciousness in zero tokens
• FAULT-TOLERANCE.md — Ternary degradation is graceful by default
• MEMORY-ARCHITECTURE.md — Five-tier memory topology from GPU registers to persistent stores
• MUSCLE-MEMORY.md — Proprioception for agents: chord shapes, flex, recall
• TRIPARTITE-SYNC.md — HARDCODE/MODEL/HYBRID/CACHED: when to think, when to act
• CONTEXT-WINDOW-ECONOMICS.md — Attention is the scarce resource
• AGENT-TO-AGENT-PROTOCOL.md — Ternary signals replace message passing

• EDGE-CASES-AND-BOUNDARIES.md — Where Z₃ gets interesting
• THE-PACKED-FORMAT.md — How 20 trits fit in 32 bits, the 2-bit encoding
• TERNARY-NUMBERS.md — Why {-1, 0, +1} is the most powerful three symbols
• TERNARY-QUANTIZATION.md — Quantizing neural networks to {-1, 0, +1} weights
• SPARSITY-IN-TERNARY-SYSTEMS.md — Why zero is a first-class symbol, not truncation
• FLEET-MAP.md — 303 crates, their relationships, and how to navigate them
• CRATE-PATTERNS.md — Every crate follows one of 7 patterns. Learn them.
• TESTING-AS-PROOF.md — The test suite is a theorem prover with 5,300 lemmas

• ESP32-AS-BODY.md — The microcontroller is the agent's hand
• EMBODIED-AI.md — Intelligence distributed across body, GPU, and host
• GPU-AS-MOTOR-CORTEX.md — Ternary on metal: 16x denser than FP32
• FLUX-TO-PTX.md — The compilation pipeline from intent to silicon
• DISTRIBUTED-COMPUTE-GRAPH.md — Execution graphs that span agents and failure domains

• VISUALIZING-TERNARY-DATA.md — Making {-1, 0, +1} visible
• COST-ECONOMICS.md — The dollar math of ternary computing
• CONSTRUCT-ECONOMICS.md — Markets, pricing, and value flow for agent-generated capabilities
• FAILURE-MODES.md — A catalog of how ternary systems fail
• TROUBLESHOOTING.md — When the ternary world does not make sense
• THE-CLI-REFERENCE.md — Command-line tools for agent operations
• MULTI-MODALITY.md — Thinking in vision, audio, text, and proprioception
• INTEGRATING-WITH-EXISTING-ML.md — Ternary weights in PyTorch and TensorFlow
• MIGRATING-TO-TERNARY.md — Porting binary code to {-1, 0, +1}
• SECURITY-MODEL.md — Ternary permissions and threat topology
• BENCHMARKING-TERNARY.md — Reproducible methodology for the 16× claim
• DEBUGGING-AND-TRACING.md — Following intent through five layers
• OBSERVABILITY-AND-TELEMETRY.md — Monitoring a ternary fleet as a field of signals
• HOW-TO-FLEX.md — Using openmind's muscle memory API
• HOW-TO-INGEST.md — Making any codebase into chord shapes
• HOW-TO-CONDUCT.md — Orchestrating multiple agents with shared muscle memory
• REASONING-AND-PLANNING.md — The layered stack of agent cognition
• ONLINE-LEARNING.md — Continuous improvement from the stream of experience
• HOW-TO-EXTEND.md — Adding your own crates to the fleet

• THE-CHAIN-REACTION-PATTERN.md — The HOOK/REVEAL/CONNECT/ACTIVATE structure of every document
• THE-IMPOSSIBILITY-PROOF.md — What ternary cannot do
• AGENT-SELF-ASSESSMENT.md — How to audit your own understanding
• ANTI-PATTERNS.md — What not to do in ternary systems
• THE-MANIFESTO.md — Why we build this way
• THE-COOKBOOK.md — 12 recipes for common ternary tasks
• FAQ.md — Questions agents actually ask
• GOVERNANCE-AND-COMMUNITY.md — How the ecosystem decides
• HUMAN-IN-THE-LOOP.md — Tiered human involvement based on confidence and stakes
• PRIVACY-IN-TERNARY-SYSTEMS.md — Obfuscation by structure
• THE-ROADMAP.md — Where the fleet is going
• CASE-STUDIES.md — Ternary in the wild: three real deployments
• DESIGN-PRINCIPLES.md — The seven rules that govern every decision
• PERFORMANCE-OPTIMIZATION.md — Making ternary code fly
• RESEARCH-METHODOLOGY.md — How new ideas enter the ecosystem
• GLOSSARY.md — 23 terms that unlock the entire ecosystem
• VERSIONING-AND-COMPATIBILITY.md — 303 crates, one moving target
• LOOM-TAPESTRY.md — Why documentation is the deepest engineering
• FLUID-CHAINS.md — Every node decides; model↔code boundary dissolves
• CELLULAR-JUPYTER.md — Jupyter as a cellular computation fabric

• Any AI agent encountering SuperInstance repos for the first time
• Agent frameworks looking for a2a documentation patterns
• Humans who want to see what agent-native docs look like

If an agent reads one page and doesn't change how it thinks about the system, that page failed. Every document must rewire understanding.

Apache-2.0