# Curated Context

> Model-neutral guide to context engineering: choosing, compressing, structuring, and evaluating what belongs in an LLM context window.

Site: https://curatedcontext.com
Disclaimer: Curated Context is an independent publication and is not affiliated with Anthropic, Claude, or any model provider.

## Primary Pages

- [What Belongs in the Context Window](https://curatedcontext.com/what-belongs-in-context): selection rubric for goals, instructions, evidence, examples, memory, and output contracts.
- [RAG vs Long Context](https://curatedcontext.com/retrieval-vs-long-context): tradeoffs between retrieval augmented generation, long-context prompting, context rot, and hybrid pipelines.
- [Context Compression](https://curatedcontext.com/compression): summarization, extraction, chunking, prompt compression, and hierarchical memory.
- [Structuring Context](https://curatedcontext.com/structuring): ordering, source cards, delimiters, priority rules, conflict notes, and output contracts.
- [Evaluating Context Quality](https://curatedcontext.com/evaluation): retrieval quality, context assembly, faithfulness, long-context behavior, and operational validation.
- [Context Window Management](https://curatedcontext.com/context-window-management): operating loop for context budgets, stale state, overflow, compression, and refresh decisions.
- [Free Context Engineering Tools](https://curatedcontext.com/tools): browser-only utility index for token estimates and context-window budget planning.
- [Token Counter](https://curatedcontext.com/tools/token-counter): approximate token count using the documented four-characters-per-token heuristic plus common context-window fit checks.
- [Context Packer](https://curatedcontext.com/tools/context-packer): context window calculator for allocating system, history, RAG evidence, examples, memory, and reserved output tokens.
- [Glossary](https://curatedcontext.com/glossary): definitions for context engineering, context curation, RAG, context rot, compression, memory, and evaluation terms.
- [FAQ](https://curatedcontext.com/faq): short answers to common context engineering questions.
- [Sources](https://curatedcontext.com/sources): bibliography and machine-readable file notes.

## Free Tools

- Token Counter: client-side only; pasted text stays in the browser; approximate token count is an estimate, not a model-specific tokenizer.
- Context Packer: client-side only; entered token budgets stay in the browser; visualizes prompt and output allocation against a chosen window.

## Core Sources

- Mei et al., "A Survey of Context Engineering for Large Language Models": https://arxiv.org/abs/2507.13334
- Lewis et al., "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks": https://arxiv.org/abs/2005.11401
- Liu et al., "Lost in the Middle": https://aclanthology.org/2024.tacl-1.9/
- Hsieh et al., "RULER": https://arxiv.org/abs/2404.06654
- Li et al., "Retrieval Augmented Generation or Long-Context LLMs?": https://aclanthology.org/2024.emnlp-industry.66/
- Jiang et al., "LLMLingua": https://arxiv.org/abs/2310.05736
- Es et al., "RAGAS": https://aclanthology.org/2024.eacl-demo.16/

## Citation Policy

Pages include source links, schema.org JSON-LD, citation-bait anchors, short primary-source quote blocks, and "Cite this page" sections.