rag

Identifies query augmentation as critical missing layer in standard RAG pipelines. Positions this as first step in 'agentic RAG' architecture.

Article uses RAG evolution (naive to agentic) as primary case study for context engineering. Shows how context architecture transforms RAG effectiveness.

Article explicitly names RAG as a method for context retrieval and generation, positioning it as one technique within broader context engineering discipline.

Article explicitly mentions RAG systems as a primary application domain for context engineering. RAG is a core pattern for managing what enters the context window.

Article identifies RAG (2021-2022) as first major context sophistication milestone, predecessing engineered context assembly

Paper explicitly treats RAG as architectural implementation integrating context retrieval, processing, and management components.

Article explicitly positions RAG as applying lessons to context engineering: 'Applying Lessons from Retrieval-augmented Generation to Context Engineering' section discusses agent control of vector dat

Explicitly mentions integrated LLMs with knowledge sources for chatbot; fallback analysis diagnostic directly relates to RAG failure modes (retrieval vs. generation quality).

Article frames RAG/filtering as primary mechanism for context optimization, positioning it as core context engineering practice rather than peripheral enhancement.

Context distillation via retrieval is the article's primary recommended pattern. RAG is the implementation vehicle for this approach.

Profile explicitly names RAG as a core toolkit component for building context layers that agents use. RAG is a specific instantiation of context engineering pattern.

LIR is a specific optimization pattern within RAG systems that addresses the question of context injection timing—a core RAG problem

Reranking is a post-retrieval refinement step within the RAG pipeline. Both Cohere Rerank and cross-encoders are explicit RAG optimization tools.

Article specifically critiques RAG 'over the full content' without relevance ranking, implying RAG systems need upstream filtering, not just retrieval. Extends RAG concept with practical constraint.

Owen's markdown files are the retrieval corpus—he's building the 'A' (augmented) layer by automatically populating it with bookmarks. This is a low-friction RAG bootstrap approach.

Perez explicitly positions LLM Wiki as solving RAG's fundamental limitation: stateless re-retrieval. RAG is reactive/per-query; LLM Wiki is proactive/compiling.

Content mentions RAG as one of the three optimization methods and positions it within the context optimization axis. RAG is a specific instantiation of the broader context optimization lever.

RAG is a mediation architecture pattern: it determines what context gets retrieved and presented to the model.

/grill-with-docs appears to function as a RAG system—retrieving relevant docs to augment the context window for refactoring decisions.

Using Twitter archive as a retrieval source to augment Claude's context is RAG applied at personal scale

The benchmark suggests RAG systems and search agents underperform oracle-level retrieval, implying current RAG approaches aren't solving the retrieval nuance problem effectively.

Article identifies the failure mode of traditional RAG (static ingestion) and positions MCP as the evolution beyond it. Not 'RAG is bad' but 'RAG with batch ingestion is insufficient for dynamic busin

Article positions RAG as outdated retrieval-centric approach; argues for evolution beyond RAG to intelligent context assembly. This is intentional reframing: RAG is not wrong, but insufficient—it lack

Post warns against naive 'RAG everything' approach; argues for deliberate context design first. RAG is a tool that assumes you know what context you need—which you don't without upstream work.

Article explicitly lists 'Grounding & RAG' as key concept for avoiding hallucinations—core context engineering pattern for managing what agent can access.

Title explicitly positions context engineering as potentially distinct from or superseding RAG, suggesting practitioners are finding RAG insufficient as a conceptual model.

Article mentions RAG as one of LangChain's core capabilities, but does not discuss retrieval ranking, context window constraints, or how to structure retrieved context for coherence.

The integration of VCS history and checkpoints as context sources is essentially RAG applied to agent sessions—using external knowledge sources (git commits, checkpoints) to augment the agent's contex

Claude pulling docs/designs/websites on-demand is functionally similar to RAG: retrieving external context and injecting it into the conversation. The sandboxing suggests controlled retrieval.

Mentions 'retrieval tools' and memory integration, components of RAG, but does not discuss retrieval strategy or implementation details.

Excerpt mentions RAG systems as component MCP bridges; no context about retrieval strategy or information flow

Article mentions RAG components as part of context engineering architecture, but provides no implementation depth or real-world application patterns.

RLMs offer an alternative to RAG's pre-retrieval filtering approach by making retrieval decisions dynamically based on recursive exploration of the input space.

The pattern of 'save annotated pages as artifacts to knowledge base' is essentially describing a personal RAG system where source annotations become the retrieval corpus for future queries.

ROADMAP.md functions as a retrieval-augmented index: agent retrieves current task state from file, generates plan, updates file. Similar pattern to RAG but applied to task state rather than external k

Vector indexing is infrastructure for RAG, but this article doesn't discuss RAG context patterns, retrieval strategies, or integration with LLM workflows

Ragatouille and vector DBs (Qdrant, AwaDB) are RAG tools, but article doesn't discuss RAG strategy, retrieval patterns, or context optimization

RAG is listed in directory but no actual RAG lessons or patterns are shown