context-engineering-tools

This project is a large curated list, not a piece of software you install. It catalogs more than 160 tools related to context engineering, which the README defines as the practice of choosing, organizing, budgeting, and delivering all the information an AI language model sees when it answers a question, including system instructions, past messages, retrieved documents, examples, memory, and tool results. The authors are careful to explain that this is broader than prompt engineering, which is just about writing instructions, and broader than retrieval augmented generation, which is just about pulling in outside documents. Both are treated as pieces of the larger context engineering picture rather than the whole thing.

The list is organized into fourteen categories covering things like managing and versioning prompts, assembling context from multiple sources, compressing long histories, routing very long inputs, giving AI agents memory across sessions, retrieval, parsing and chunking documents, connecting to tools through the Model Context Protocol, counting and budgeting tokens, caching, structured output formats, debugging what context was actually sent to a model, evaluating context quality, and protecting context from security risks like prompt injection.

Each entry links back to an original source such as an official repository or product page, and tools are marked with icons showing whether they are open source, open weights, open core, or fully commercial and closed. The project also keeps a machine readable version of the same data as JSON and CSV files, and states it is reviewed monthly to keep entries current, with a note about a discontinued and historical tools section for projects that no longer apply.

The project is released under the CC0 license, which places it in the public domain with no restrictions, and it welcomes pull requests from contributors who want to add or update entries.