← Field Guide  ·  Tier 3 · Community & Curated Catalogues

Community Curations

Awesome lists · blog syntheses · cognitive-architecture papers

Community5 patterns
In Plain English

This page shows every pattern that Community Curations has published or written about — 5 in total. Each one includes a simple picture of how it works, a short explanation, an example of when to use it, and its trade-offs (the good parts and the not-so-good parts).

Reading patterns from one source at a time matters because it lets you see how a single company or team thinks about building AI agents, and compare their ideas side by side. If a pattern here also appears in the bigger Deep Catalogue, you will see a link so you can read the full, detailed version.

An open reference book with a pencil in the fold under warm lamp light

3D.01Chain-of-Thought

CHAIN-OF-THOUGHT Question Step 1 Step 2 Step 3 Answer

Prompt the model to show intermediate reasoning before the answer. Wei et al. 2022 — the oldest of modern reasoning patterns.

Use caseArithmetic, commonsense reasoning, any step-heavy task.
Trade-offsAccuracy lift · Longer outputs.

3D.02Self-Ask

SELF-ASK Agent Main Q Follow-up Q Search Answer Sub-Qneed?asksearchanswer

Agent decomposes a question into follow-up questions it asks itself, often combined with search.

Use caseMulti-hop Q&A requiring decomposition.
Trade-offsHandles multi-hop well · More turns per query.

3D.03Program-Aided Language Model (PAL)

PAL Problem Write Code Execute Use Result

LLM writes code (e.g. Python) to solve a problem; code runs; result used. Gao et al. 2022.

Use caseMath, symbolic manipulation, exact computation.
Trade-offsExact answers for computable problems · Requires execution environment.

3D.04Sparse Priming Representations

SPARSE PRIMING Full Memory Compress Priming Tokens Agent Context

Compress long memory into minimal token-efficient prompts that trigger recall. Community pattern.

Use caseContext-window-constrained agents with large memory stores.
Trade-offsToken efficiency · Fidelity loss on compression.

3D.05Cognitive Architectures (CoALA)

CoALA Memoryworking + long-term Action Spaceexternal + internal Decision Procedurepropose + evaluate

Sumers et al. (2023) framework organising LLM agents into memory + action space + decision procedure.

Use caseStructured agent design with explicit cognitive primitives.
Trade-offsClean theoretical frame · Abstract; not a framework.
Also in Tier 3
LangChain / LangGraph 5Hugging Face 3Survey Papers & Taxonomies 4