Architecture: Techniques Corpus

10 documents filed under architecture. Every one of them is retrievable by agents through rag_context. All categories.

• Business Logic Placement & Model-Level Architecture

  A visual and narrative explanation of where business logic belongs across domain, application, and infrastructure layers.

• Consensus Algorithms: Distributed Agreement Under Failure

  Algorithms and theory for achieving agreement among distributed processes despite failures: FLP, Paxos, Raft, and Byzantine consensus.

• Consistency Models: The Spectrum of Distributed Guarantees

  Formal specifications of correctness for concurrent and distributed systems, from strong to eventual consistency.

• Coordination Models: Decoupled Concurrent System Composition

  Abstract coordination mechanisms (tuple spaces, Petri nets, coordination languages) that decouple computation from communication.

• Enterprise Django: Layout, Service Layer, ORM, Migrations, APIs, Async, Tenancy, and Testing

  Decision rules for large Django codebases on the 6.1 line: app boundaries and settings split, services versus fat models, ORM and migration technique at scale, DRF versus django-ninja, async and the 6.0 Tasks framework, multi-tenancy, and test structure.

  Researched 2026-09-18 · 46 sources

• Local-First Sync Engines and CRDTs in Practice: Which Architecture, Which Engine, and What Bites in Production

  State of practice as of 2026-10 for offline-capable, multi-device sync. With a central server, CRDTs are an optimization rather than a requirement; the engines that reached GA (Electric 1.0, Zero 1.0, PowerSync) are all server-authoritative and differ mainly on the write path; CRDT libraries (Automerge 3, Yjs 13, Loro 1.x) fixed memory and load time in 2024 to 2025 and are best embedded per field; and several once-popular options (Replicache, Triplit, InstantDB cloud, cr-sqlite, YSwift) are in maintenance, sunset or stale.

  Researched 2026-10-05 · 89 sources

• Multi-Agent LLM Systems in Practice: Comms, Lifecycle, Delegation, Verification

  Evidence-audited state of multi-agent LLM systems as of Aug 2026 — what A2A/MCP/Managed Agents actually standardize, why every production stack is hub-and-spoke with caps, the delegation-as-contract discipline, the honest data on code-graph tooling vs grep, and why verification must be held-out and execution-based. Separates what the field agrees on from what is still folklore.

  Researched 2026-08-14 · 24 sources

• MX Rust CLI and MCP Server Development Guide

  A comprehensive guide for building, managing, and extending the MechCrate Rust CLI (mx) and MCP server.

• Project & File-Structure Theory for Scalable Applications

  Principles and patterns for structuring application folders and files so they scale with codebase, team, and feature complexity.

• System One Decision Models: Jev, Typed Questions, and Calibrated Routing Around LLMs

  What a System One model is, what Jev actually returns (typed answers plus calibrated probabilities, no generated text), the patterns its docs prescribe, where it fits around an LLM, and what the independent evidence three days after launch does and does not support.

  Researched 2026-09-18 · 42 sources