Overview

What is fortemi-react?

Fortémi is an intelligent database for AI-ready applications: normalized data contracts plus data-science and processing tooling that turn messy organizational data into searchable, linkable, provenance-aware structures. The Fortemi server owns the live APIs and Knowledge Shard authority; `@fortemi/core` consumes declared browser profiles, and HotM consumes the server and Core contracts as an application.

fortemi-react is the browser edition of that stack. It runs a normalized PGlite data layer with no separate server and exchanges state through named, receipt-bound profiles. The data lives in the browser on the user's device, search runs locally, and any cloud or AI provider is strictly opt-in. Keeping notes for an AI agent and searching them by meaning, not just exact words, is a use case built on that substrate — not the whole product.

It is the stack in the browser. (HotM, a separate project, is a desktop client that bundles the Rust server; fortemi-react is the stack itself, ported to run in-page.) Knowledge Shard exchange is governed by named compatibility profiles and cross-repository conformance tests; repository and API shapes outside a selected profile are not implied to be identical.

Fortémi (product) vs `@fortemi/core` (embeddable library)

`@fortemi/core` is an embeddable library — the headless data layer other tools build on. (For example, it is AIWG's default index and discovery backend, reading from a static local cache.) A tool that embeds `@fortemi/core` is not running Fortémi: it is using the library's schema and query surface inside its own process, and the library ships no data anywhere by default. Fortémi the product is the full stack — the server or this browser edition — with its job pipeline, capability system, and portable archives.

Highlights

• Runs fully in-browser — a real PostgreSQL data layer via PGlite (PostgreSQL compiled to WebAssembly) with pgvector. No backend to stand up.
• Hybrid search — full-text and semantic (meaning-based) retrieval, fused with rank fusion, running off the main thread so the page never freezes.
• Three ways to load notes — an in-browser database, plain files with no database, or a ready-made snapshot that starts fast. One shared design picks the best source for what you asked.
• Local-first, private-by-default — your notes stay on the device; nothing leaves it unless you choose a remote or cloud provider.
• Profile-scoped portability — Knowledge Shards validate before import and move between supported consumers only for their declared, conformance-tested profile.

Supported contract evidence

Fortemi run 6393 passed the declared authority -> React/Core -> HotM consumer surface on Linux x86_64, Linux arm64, and macOS arm64 with `@fortemi/[email protected]` at commit `ccf96fad6025025293e40e250c85f088c8999d86`. The Knowledge Shard claim is exactly `2.0.0/full-v1`. Windows is a separate deferred authority story in Fortemi #1096. The parent audit remains `NO-GO`; the run does not establish suite-wide portability, complete backup, GUI/native-dialog coverage, or every platform and architecture.

Three ways to load your notes

fortemi-react reads your notes through one shared contract, no matter where they live. A "picker" chooses the source that fully answers your request with the least startup cost:

Source	What it is	Best for
In-browser database	A full PGlite database with pgvector	Ranked search and vectors on read/write archives
Canonical records	A lightweight writable record store — no database	Read/write note-taking with instant startup
Plain files	A knowledge pack read in place as static files — no database	Read-only reference sets on any static host
Snapshot	A database with its search index already built in	Fast startup on large sets

Same code, your choice, no rewrite. Small sets can read and write with no database at all; the full database layers in ranked full-text and semantic search when you need it — and it can be rebuilt from the canonical records at any time.

How it differs from typical browser storage

Aspect	Typical browser storage	fortemi-react
Store	`localStorage` / IndexedDB blobs	PGlite — a real PostgreSQL engine with pgvector
Search	Substring match	Hybrid full-text + semantic, fused with RRF
Server	Needed for real search	None — everything runs in the page
AI / cloud	Baked in	Opt-in, private-by-default
Portability	App-specific	Knowledge Shards with named, conformance-tested profiles
Load modes	One	Three — database, plain files, or snapshot

The packages

fortemi-react is a small monorepo of three published packages, in a linear dependency chain:

• `@fortemi/core` — the headless, browser-only data layer: PGlite, a single-writer worker, MCP tools, a job queue, and hybrid search. No React.
• `@fortemi/graph` — framework-agnostic community-graph tooling: layout, filtering, neighborhood expansion, and snapshot serialization. Depends on core; no React.
• `@fortemi/react` — React 19 hooks and `FortemiProvider` over `@fortemi/core`, with a `GraphView` built on `@fortemi/graph`.

A plain JavaScript app can use `@fortemi/core` and `@fortemi/graph` with no React at all.

Where to go next

• Getting Started — install the packages and wrap your app in `FortemiProvider`.
• Integration — add fortemi-react to an existing React app.
• API Reference — every hook, provider, and core API.
• Packages — per-package reference for core, graph, and react.
• Extending — add custom capabilities and deploy.
• Blog — monthly release reports and what's new.