https://github.com/manifesto-ai/core
Semantic layer for deterministic domain state
https://github.com/manifesto-ai/core
agent ai declarative deterministic neuro-symbolic neuro-symbolic-ai semantic
Last synced: 28 days ago
JSON representation
Semantic layer for deterministic domain state
- Host: GitHub
- URL: https://github.com/manifesto-ai/core
- Owner: manifesto-ai
- License: mit
- Created: 2025-12-30T23:07:09.000Z (5 months ago)
- Default Branch: main
- Last Pushed: 2026-04-23T04:35:20.000Z (about 1 month ago)
- Last Synced: 2026-04-23T06:27:58.802Z (about 1 month ago)
- Topics: agent, ai, declarative, deterministic, neuro-symbolic, neuro-symbolic-ai, semantic
- Language: TypeScript
- Homepage: https://manifesto-ai.dev
- Size: 9.89 MB
- Stars: 14
- Watchers: 0
- Forks: 0
- Open Issues: 14
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- Contributing: CONTRIBUTING.md
- Funding: .github/FUNDING.yml
- License: LICENSE
- Security: SECURITY.md
- Agents: AGENTS.md
Awesome Lists containing this project
README
# Manifesto
**Semantic Layer for Deterministic Domain State**
Manifesto gives you one semantic model for deterministic domain state and tooling surfaces built from the same schema. Approval, history, and governance come later only when the project needs them.
[](https://www.npmjs.com/package/@manifesto-ai/core)
[](LICENSE)
## Install
```bash
npx @manifesto-ai/cli init
```
That opens the interactive init flow. Prefer pnpm or bun? Use `pnpm dlx @manifesto-ai/cli init` or `bunx @manifesto-ai/cli init`.
Prefer manual setup? Install `@manifesto-ai/sdk` for the app runtime. Install `@manifesto-ai/compiler` directly only when your project imports compiler entrypoints such as `@manifesto-ai/compiler/vite`.
## Quick Example
```mel
domain Counter {
state { count: number = 0 }
action increment() {
onceIntent { patch count = add(count, 1) }
}
}
```
```typescript
import { createManifesto } from "@manifesto-ai/sdk";
import CounterMel from "./counter.mel";
const app = createManifesto(CounterMel, {}).activate();
await app.dispatchAsync(
app.createIntent(app.MEL.actions.increment),
);
console.log(app.getSnapshot().data.count); // 1
```
## Start With The Docs
- Build the first app: [Docs Home](./docs/index.md) -> [Guide Introduction](./docs/guide/introduction.md) -> [Quick Start](./docs/guide/quick-start.md)
- Set up CLI, editor, AI, or Studio workflows: [Developer Tooling](./docs/guides/developer-tooling.md)
- Add approval, review, or sealed history later: [When You Need Approval or History](./docs/guides/approval-and-history.md)
- Look up a package you already know: [API Reference](./docs/api/index.md)
- Go deeper into the model: [Concepts](./docs/concepts/index.md), [Architecture](./docs/architecture/index.md), [Internals](./docs/internals/index.md)
Start with the base SDK runtime. Add compiler entrypoints, Lineage, Governance, or the surrounding DX packages only when the project actually needs them.
## What Manifesto Is Not
- Not a state management library
- Not an AI framework
- Not a database or ORM
- Not a workflow engine
## Research & Citation
Manifesto is a general-purpose declarative runtime. The underlying design philosophy and the initial empirical validation of its reflective protocol
If you use this core framework in your research, please cite:
```
@misc{jung2026llmdoesselfrevisingagent,
title={How Much LLM Does a Self-Revising Agent Actually Need?},
author={Sungwoo Jung and Seonil Son},
year={2026},
eprint={2604.07236},
archivePrefix={arXiv},
primaryClass={cs.AI},
url={https://arxiv.org/abs/2604.07236},
}
```