# Encyclopedia of Abstractions

> A cross-domain catalog of recurring structural patterns and the interventions that address them: 1,524 prime abstractions, 11,102 domain-specific abstractions, 1,134 solution archetypes and 9,939 mechanisms, joined by 12,854 typed relationships, trigger logic on every archetype, and a registry of 33,381 cited works. Queryable locally through an MCP server; every catalog downloadable as data. Experimental, AI-assisted, and revised continuously.

Prime abstractions are patterns that recur across at least three fields (feedback, equilibrium, trade-off). Domain-specific abstractions are their field-bound relatives. Solution archetypes are recurring shapes of intervention, and mechanisms are concrete ways to carry them out. Read the full briefing before using the catalog for a task.

## Start here

- [Full briefing](https://abstractopedia.org/llms-full.txt): what the site contains, the data model, how to use it for a task, machine access, and its limits, in one file
- [The same briefing as a page](https://abstractopedia.org/for-ai/)
- [About](https://abstractopedia.org/about/): what the project is and is not

## Catalogs

- [Prime abstractions](https://abstractopedia.org/primes/): 1,524 cross-domain patterns
- [Domain-specific abstractions](https://abstractopedia.org/domain-specific/): 11,102 field-bound patterns
- [Solution archetypes](https://abstractopedia.org/archetypes/): 1,134 recurring interventions, each with trigger logic
- [Mechanisms](https://abstractopedia.org/mechanisms/): 9,939 concrete implementations of archetypes

## Structure and logic

- [Typed hierarchy: consumer contract](https://abstractopedia.org/conceptual/typed-mixed-dag-consumer-contract/): what each relationship type claims
- [Relationships](https://abstractopedia.org/relationships/): counts by type and qualifier
- [Archetype trigger logic: method](https://abstractopedia.org/conceptual/from-problem-descriptions-to-defensible-abstraction-links/): what a grounding claims and what it does not
- [Abstraction Atlas](https://abstractopedia.org/atlas/) and [Shape of the abstraction space](https://abstractopedia.org/shape/): maps of the catalog

## Machine access

- [MCP server](https://abstractopedia.org/mcp/): local, read-only, 48 tools; install with `curl -fsSL https://abstractopedia.org/downloads/eoa-mcp-install.sh | bash`
- [Downloads](https://abstractopedia.org/downloads/): every catalog and graph as JSON/JSONL
- [Checksum manifest](https://abstractopedia.org/downloads/downloads.manifest.json): every registered download with its size and SHA-256

## References, learning, research

- [Reference registry](https://abstractopedia.org/references/): 33,381 cited works
- [How references were verified](https://abstractopedia.org/references/methodology/): support grades and what they establish
- [Learnability curriculum](https://abstractopedia.org/learnability/): the primes ordered easiest first, with reading levels
- [Applications and experiments](https://abstractopedia.org/applications/), including [a null result](https://abstractopedia.org/conceptual/retrospective-testing-runtime-abstraction-scaffolding/) on runtime scaffolding of frontier models
- [Conceptual essays](https://abstractopedia.org/conceptual/)

## Limits

An experimental, AI-assisted synthesis by one author, not a peer-reviewed reference; it contains errors, and entries are revised as they are found. 32% of citations carry a published grade from checking against their sources so far; checking sources, internal consistency and arithmetic has corrected 1,132 errors. One published experiment found that scaffolding a frontier model's reasoning with the encyclopedia at inference time was largely inert on self-generated problems within the model's reach; early, unpublished observations suggest it may help on much harder problems, and follow-up experiments are pending.