A memory layer that learns. Instead of storing and retrieving, Adapt observes incoming data, builds understanding, and reshapes its own structure over time. It answers questions that databases and RAG pipelines can't — the ones that require paying attention as data flows in.

Documentation · Changelog · Releases · npm · Issues

Install

npm install @unbody-io/adapt

Adapt uses the Vercel AI SDK for LLM access. Install a provider:

npm install @ai-sdk/openai    # or @ai-sdk/anthropic, @ai-sdk/google, etc.

Quick start

import { Brain } from '@unbody-io/adapt'
import { openai } from '@ai-sdk/openai'

const brain = new Brain({
  prompt: 'Track my coding patterns and development philosophy.',
  model: openai('gpt-4o'),
})

await brain.inject([
  { type: 'commit', message: 'refactor: extract validation into pure functions' },
  { type: 'review', comment: 'Too heavy — factory functions work fine for our scale.' },
])

const result = await brain.ask('What is my coding philosophy?')
console.log(result.insight)

Features

• Self-evolving — creates, merges, splits, and removes Neurons based on usage
• Any LLM — cloud or local via Vercel AI SDK
• Pluggable stores — in-memory or SQLite
• Modular — use the Brain orchestrator or individual Neurons standalone

Limitations

• Requires models with structured output and tool calling support
• Local model support (Ollama, LMStudio) not fully tested yet
• Not a database — builds understanding, doesn't store raw data
• Experimental (0.0.x) — expect breaking changes

Contributing

Found a bug or have an idea? Open an issue. PRs welcome.

MIT — Unbody