brain_kernel 0.1.4

Headless system kernel for knowledge-grounded multi-agent systems. Bundles project / canonical / patch / validate / build / MCP / chat / RAG over mcp_bundle and flowbrain_core. Products (builder / ind [...]

brain_kernel

Headless system kernel for knowledge-grounded multi-agent systems.

brain_kernel bundles project / canonical / patch / validate / build / MCP / chat / RAG over mcp_bundle and flowbrain_core, and exposes them as a single Dart library that products (builders, industrial HMIs, medical / education tools, B2B platforms, personal apps) can wire into their own UI and domain workflow without writing the integration plumbing themselves.

What's inside

• Core — project / canonical / patch pipeline / asset validator / undo-redo stack / sidecar logs (prefs, chat, history, undo).
• Feature — BM25 index, gold-question runner, asset extractor + reviewer queue, asset-touch observer.
• Infra — bundle reader / knowledge writer / mcpb packager, embedding runner, BM25 query engine + bundle registry, domain storage, FlowBrain wiring (KvStoragePort adapter, runtime probe, LLM port adapter, FlowDefinitionWorkflow), MCP server bootstrap (tool scope + transport picker), LLM session manager, chat controller + system-prompt composer.
• System — BundleActivation + BundleActivationRegistry: the single standard API every host (AppPlayer Studio · AppPlayer · future hosts) uses to activate a bundle, register its assets, and tear it down. Per-bundle isolation via <bundleId>.<asset.id> prefixing.

Quick start

import 'package:brain_kernel/brain_kernel.dart' as bk;

// Boot the kernel.
final wiring = bk.FlowBrainWiring(
  workspaceId: 'my_workspace',
  kvStoragePort: bk.InMemoryKvStoragePort(),
);
await wiring.boot();

// Activate a bundle.
final activation = bk.BundleActivation(
  system: wiring.system,
  bundleId: 'my.bundle',
);
await activation.activate(myBundle); // McpBundle

Hosts (AppPlayer Core, AppPlayer Studio, ...) drive this lifecycle from their own session-management code.

Re-exports

The barrel re-exports the upstream packages so consumers can stay on a single MCP surface (per FR-CMP-002):

• flowbrain_core — KnowledgeSystem, the five facades, the agent runtime types, infrastructure ports.
• mcp_bundle — bundle schema, validators, ports.
• mcp_server — MCP server primitives.
• mcp_client — selected client types (Client, transport configs).

Products do not need to depend on these packages directly; depending on brain_kernel is enough.

Status

brain_kernel is the canonical kernel for the MakeMind ecosystem and the foundation that AppPlayer Core, AppPlayer Studio, and the FlowBrain products build on. The API surface is stable enough for early adopters but still evolves with the upstream mcp_bundle / flowbrain_core spec — pin caret versions.

License

MIT