Topic · topic
Context Engineering
Context engineering is the practice of curating the information that large language models receive at inference time so that the model can perform a task reliably and cost-effectively. It treats the context window as a finite attention budget and looks for the smallest set of high-signal tokens that maximize the likelihood of the desired outcome. Context engineering subsumes and extends prompt engineering, system prompts, tool design, retrieval, agent loops, structured note taking, compaction, and multi-agent decomposition. It is a foundational discipline for building production AI agents and assistants.
Resources
-
Effective Context Engineering for AI Agents
Anthropic's engineering guide to context engineering, framing context as a finite attention budget and walking through system prompts, tool design, few-shot examples, just-in-time retrieval, compaction, structured note taking, and multi-ag…
-
Retrieval-Augmented Generation (RAG)
RAG is a context engineering pattern that augments LLM prompts with passages retrieved at inference time from a vector store, search index, or knowledge base. RAG keeps facts outside the model and is one of the most widely used context eng…
-
Prompt Engineering
Prompt engineering is the discipline of crafting model instructions and examples to guide model behavior. Prompt engineering remains a sub-discipline of context engineering and includes techniques like role prompting, chain-of-thought, few…
-
Agentic Loops and Tool Use
Agentic loops are iterative reasoning patterns in which an LLM plans, calls tools, observes results, and refines its plan. Tool design is a central context engineering concern: tools must be token-efficient, have minimal overlap, and inclu…
-
Long-Horizon Context Strategies
Long-horizon strategies handle conversations and tasks that exceed the context window. Techniques include compaction (summarizing history into a smaller representation), structured note taking (persistent external memory), and multi-agent…
Links
Providers working in Context Engineering
Providers whose own tags share at least two of this topic's tags, most shared first — the top 30 of 558.
| Provider | About | Rating | APIs |
|---|---|---|---|
| Cognee | Cognee is an open-source AI memory and knowledge graph platform that enables developers to build persistent, structured memory for AI agents and LLM applications. The platform provides a REST API and Python/TypeScript SDKs for ingesting do… | developing | 1 |
| Anthropic | Anthropic is an AI safety company and the creator of the Claude family of large language models (Opus, Sonnet, Haiku, and the Fable/Mythos frontier line). The Claude Developer Platform exposes them through a single REST API at api.anthropi… | exemplar | 6 |
| Dust | Dust is a Paris-based enterprise AI platform for building, deploying, and operating teams of AI agents that have shared context across a company's knowledge and tools. Dust positions itself as the platform for "AI Operators" — the people w… | exemplar | 9 |
| Seekr | Seekr Technologies builds explainable, auditable, sovereign AI for regulated industries and high-stakes government missions. Its platform, SeekrFlow, is an end-to-end AI operating system that covers document ingestion and AI-ready data pre… | exemplar | 8 |
| Compresr | Compresr is an LLM context-compression API. You send the long context you would otherwise pass to a model plus the query you want answered, and Compresr returns a shorter context that keeps the answer-bearing spans and drops the rest — few… | strong | 1 |
| Amazon Bedrock | Amazon Bedrock is a fully managed AWS service that makes high-performing foundation models from leading AI companies available through a unified API for building generative AI applications. It supports text and image generation, conversati… | strong | 2 |
| PydanticAI | PydanticAI is an open-source, model-agnostic Python agent framework built by the Pydantic team, designed to bring the ergonomic, type-safe design philosophy of FastAPI to production-grade generative AI application development. It provides… | strong | 2 |
| Letta | Letta (formerly MemGPT) is a stateful AI agents platform built around long-term memory, tool execution, and multi-agent coordination. The Letta REST API exposes 239 endpoints across 36 public resource categories — agents, memory blocks, ar… | strong | 2 |
| Vectara | Vectara is a Retrieval Augmented Generation (RAG) as a service platform that provides grounded generative AI for enterprises. The API-first platform exposes a unified REST API v2 for managing corpora, ingesting documents, performing semant… | strong | 1 |
| Flowise | Flowise is an open-source, low-code visual builder for LangChain-based LLM workflows and AI agents. Built on Node.js and TypeScript as a pnpm/Turbo monorepo, Flowise lets developers and non-developers compose chatflows, multi-agent agentfl… | developing | 1 |
| RAGFlow | RAGFlow is the open-source Retrieval-Augmented Generation engine built by InfiniFlow Inc. It combines deep document understanding (DeepDoc parsing of PDFs, images, tables and scanned files) with hybrid retrieval — dense vector search, BM25… | developing | 1 |
| Stacks Ai | StackAI (Stack AI, Inc.) is an enterprise AI agent platform that lets teams build, deploy, and govern no-code agentic workflows at scale. Its visual Workflow Builder chains LLMs, knowledge bases, connections, and logic nodes into productio… | developing | 1 |
| AI21 Labs | AI21 Labs is an enterprise foundation-model company best known for the Jamba family of open-weight hybrid Mamba/Transformer models and AI21 Maestro, a dynamic planning system that orchestrates tools, retrieval, and validated output during… | developing | 1 |
| Julep | Julep is an open-source platform for building stateful AI agents that remember past interactions and execute long-running, multi-step tasks. Its cloud API and self-hostable server expose agents, sessions, tasks, executions, documents (RAG)… | thin | 1 |
| Langbase | Langbase is a serverless AI developer platform for building, deploying, and scaling AI agents and applications. Its composable primitives - Pipes (agents), Memory (managed RAG), Threads, Agent (one API over 100+ LLMs), Tools, Parser, Chunk… | thin | 1 |
| Ondemand | OnDemand AI (on-demand.io) is a RAG-powered AI Platform-as-a-Service that lets companies infuse AI into their products without managing model infrastructure. The platform exposes a REST API for chat sessions and queries against a library o… | thin | 1 |
| Adapter | Adapter is a cognition API — a persistent memory and knowledge-graph layer that sits alongside AI models so understanding is already available when agents need answers. Founded by Adam Ghetti and David Bader, Adapter continuously reads and… | thin | 1 |
| Miriel | Miriel is the context engine and platform for AI-native development. It gives AI apps and agents the context they need in real time through a simple API: developers connect a data source with the learn operation and retrieve relevant conte… | emerging | 7 |
| Dify | Dify is an open-source platform for building AI applications, combining Backend-as-a-Service and LLMOps to streamline the development of generative AI solutions for developers and non-technical innovators alike. Teams build agentic workflo… | exemplar | 2 |
| Exa | Exa is a web search API and AI research platform built specifically for LLMs and agents — semantic and keyword search across the open web with token-efficient highlights, structured outputs, sub-200ms latency tiers, and verticals for code,… | strong | 7 |
| LlamaParse | LlamaParse is an enterprise document parsing and AI pipeline platform from LlamaIndex that converts complex PDFs, Office files, and 130+ document formats into LLM-ready structured outputs. The platform offers six composable products under… | strong | 1 |
| Edgee | Edgee is a French edge-native AI Gateway that sits between coding agents and LLM providers, intercepting, routing, compressing, metering and securing every request. Its OpenAI-compatible gateway API at edgee.io exposes chat completions, an… | strong | 1 |
| Perplexity | Perplexity AI is an answer engine that delivers accurate answers to complex questions using large language models with real-time web search capabilities. | strong | 1 |
| Serper | Serper is the world's fastest and most affordable Google Search API, delivering real-time SERP data in 1-2 seconds via a simple REST interface. It supports web search, images, news, maps, places, videos, shopping, scholar, patents, and aut… | strong | 2 |
| Scorecard | Scorecard is a simulation and evaluation platform for building, testing, and deploying frontier AI agents. Teams run their agents through thousands of realistic scenarios, judge outputs with configurable AI, human, and heuristic metrics, a… | developing | 1 |
| SambaNova Systems | SambaNova Systems is an AI infrastructure company that builds custom Reconfigurable Dataflow Unit (RDU) chips and the SambaCloud, SambaStack, and SambaRack platforms for fast, energy-efficient AI inference. Its developer-facing product, Sa… | developing | 2 |
| H Company | H Company (hcompany.ai) is a Paris-based AI lab, backed by Accel and Creandum, that builds the Holo family of vision-language models and a Computer-Use Agents platform for automating work on browsers and desktops. It ships two public APIs:… | developing | 1 |
| LinqAlpha | LinqAlpha is a domain-specialized multi-agent AI platform for institutional investment research, serving hedge funds, asset managers and investment banks. It runs retrieval-augmented, agentic analysis across SEC filings, earnings-call tran… | developing | 1 |
| Unify | Unify is an LLM routing and model gateway platform that enables developers to access 100+ large language model providers through a single unified REST API and API key. The platform dynamically routes each prompt to the optimal model based… | developing | 1 |
| Dedalus Labs | Dedalus Labs builds infrastructure for AI agents. It runs two production APIs: the Dedalus Agents API, an OpenAI-compatible MCP gateway that lets you mix and match any model from any provider with tools drawn from the Dedalus MCP marketpla… | developing | 2 |
Tags
AgentsArtificial IntelligenceAnthropicCompactionContext WindowLLMMemoryPrompt EngineeringRAGTools