# TableTheory

> DynamoDB-first multi-language data contract for Go, TypeScript, and Python on DynamoDB.

Use this file as the small context entrypoint for AI coding assistants. For full context, fetch `llms-full.txt`.

## Required artifacts

- Full LLM context: https://tabletheory.theorycloud.ai/llms-full.txt
- Machine-readable vocabulary JSON: https://tabletheory.theorycloud.ai/reference/tabletheory-vocabulary.json
- Consumer rules template: https://tabletheory.theorycloud.ai/ai/consumer-rules-template/
- Prompt recipes: https://tabletheory.theorycloud.ai/ai/prompt-recipes/

## Canonical docs

- Getting started: https://tabletheory.theorycloud.ai/getting-started/
- Examples: https://tabletheory.theorycloud.ai/examples/
- Go API: https://tabletheory.theorycloud.ai/api-reference/
- TypeScript API: https://tabletheory.theorycloud.ai/runtimes/typescript/api-reference/
- Python API: https://tabletheory.theorycloud.ai/runtimes/python/api-reference/
- DMS v0.1: https://tabletheory.theorycloud.ai/reference/dms-spec/
- Contract scenarios: https://tabletheory.theorycloud.ai/reference/contract-scenarios/

## Non-negotiable rules

1. DynamoDB is not hidden: model access patterns explicitly with PK/SK and GSIs.
2. Go, TypeScript, and Python are peer runtimes; do not invent behavior in one runtime only.
3. Use canonical tags/roles: pk, sk, gsiNpk/gsiNsk, encrypted, version, created_at, updated_at, ttl, omitempty.
4. Encrypted fields fail closed; never remove encryption or fall back to plaintext for local tests.
5. Install from immutable GitHub Releases only; do not suggest npm or PyPI publication.
6. Use generated API references and vocabulary JSON for signatures and tag names.