Wiki Index

Maintained automatically by AI. Do not edit manually.

How to Use

Tell AI:

“ingest 70 Raw/filename” → AI digests material and updates this index
“query: your question” → AI searches Wiki to answer, saves results to 85 Outputs/
“lint” → AI runs a health check on the Wiki

AI Automation Maturity L1-L5 — L1-L5 maturity model; watershed at L3 (tool proficiency vs system design)
AI Agent vs Skills vs Workflow — When to use each; the uncertainty x failure cost x token cost equation
Fine-tuning vs Prompting vs RAG — Decision framework: when to fine-tune, when to prompt, when to RAG, distillation patterns

Claude Code Architecture Deep Dive — Agent loop, 7-layer memory, 3 subagent models, KAIROS, anti-distillation, 10 commercial patterns
OpenClaw Architecture Deep Dive — PI runtime, hub-and-spoke Gateway, markdown skills, cron autonomy
DeerFlow SuperAgent Framework — ByteDance’s LangGraph-based SuperAgent with Docker sandbox, 5 agent roles, checkpointing
Mem0 Memory Architecture — LLM extraction pipeline, graph memory (Mem0-g), Mem0 vs Letta decision framework
Mastra Agent Framework — TypeScript-native agent framework with Observational Memory (94.87% LongMemEval, no vector DB)

Multi-Agent Architecture Patterns — LangGraph vs Claude SDK vs CrewAI; coordinator pattern, model tiering, production trade-offs
Agent Memory Systems — Claude Code 7-layer, Letta/MemGPT, Mem0; pointer-based architecture and autoDream consolidation
RAG Architecture Patterns 2026 — Hybrid search, GraphRAG, Agentic RAG, contextual retrieval, 3-stage pipeline

LLM Security and Guardrails — OWASP LLM Top 10, prompt injection defense, guardrail frameworks, layered defense pattern
Production Prompt Engineering — System prompt architecture, caching economics, structured outputs, meta-prompting
AI Evals and Testing — Eval-driven development, RAG testing, agent evaluation, platform comparison
LLM Inference and Serving — vLLM, quantization (AWQ/GGUF), hardware guide, self-hosted vs API crossover
AI Cost Optimization — Prompt caching, model tiering, batch API, token economics, production cost examples
Data Engineering for AI — RAG pipelines, chunking, embeddings, data quality monitoring, synthetic data, PII handling

AI Product Architecture — End-to-end system design: auth, billing, API, frontend, queues, multi-tenant, scaling
AI UX Patterns — Streaming UI, loading states, error handling, feedback mechanisms, conversation UX
AI Product Metrics — Task completion, deflection rate, CSAT, cost per conversation, OKR templates
LLM Observability and Monitoring — Production monitoring, tracing tools, OpenTelemetry, cost alerts, debugging

MCP Architecture and Ecosystem — Model Context Protocol: spec, transports, OAuth, 1000+ servers, adoption landscape
AI Coding Tools Landscape 2026 — Claude Code vs Cursor vs Copilot vs Codex vs Devin; 3 architectural camps

Open Source LLM Landscape 2026 — Gemma 4, Qwen 3.5/3.6, Llama 4, DeepSeek; benchmarks and selection guide
Ollama Local LLM Runner — Run open-source LLMs locally with OpenAI-compatible API
n8n Workflow Engine — Open-source AI-native visual workflow automation

Karpathy LLM Education Stack — MicroGPT → nanoGPT → llm.c → Autoresearch; algorithm to hardware to autonomous research