Arra Oracle - MCP Memory Layer

by Soul-Brews-Studio

TypeScript MCP server for semantic search over Oracle philosophy using SQLite FTS5 and ChromaDB hybrid search, with HTTP API and vault CLI. Requires Bun runtime (>=1.2.0).

Gaming & entertainmentstdio or Streamable HTTPCommunity

Repository-wide counts · Cached 2026-03-25

Overview

The Arra Oracle - MCP Memory Layer MCP server is a publicly available project. Review the upstream repository for installation instructions, supported tools, compatibility, permissions, and current maintenance status.

Configuration

Configuration, transport, authentication, and runtime requirements vary by project. Open the repository before connecting and use the smallest set of credentials and permissions required.

Open the Arra Oracle - MCP Memory Layer repository to read the latest documentation.

KEEP EXPLORING

Compare source, connection, and authentication details before choosing an implementation.

View the complete category

BlenderMCP

mcpblender

Community

Blender integration through the Model Context Protocol enabling prompt-driven 3D modeling, scene creation, and real-time manipulation directly from your AI assistant. Requires Blender 3.0+, Python 3.10+, and the 'uv' runtime (installed separately).

Godot MCP

Coding-Solo

Community

用于与 Godot 游戏引擎接口的 MCP 服务器。提供启动编辑器、运行项目和捕获调试输出的工具。需要在系统上安装 Godot 引擎,路径可通过环境变量 GODOT_PATH 进行配置。

MCP Unity 编辑器(游戏引擎)

CoderGamester

Community

MCP Unity 是一种适用于 Unity 编辑器的模型上下文协议(Model Context Protocol)实现,允许 AI 助手通过 Node.js 服务器桥接与您的 Unity 项目进行交互。运行该服务器需要 Unity 2022.3 及以上版本和 Node.js 18 及以上版本。

AI Game Developer — Unity MCP

IvanMurzak

Community

Unity MCP is an AI-powered game development assistant for Editor & Runtime. Connect Claude, Cursor, & Windsurf to Unity via MCP. Automate workflows, generate code, and enable AI within your games. It works inside your compiled game, allowing for real-time AI debugging and player-AI interaction.

FROM THE SOURCE

Repository README

Build-time snapshot · Retrieved 2026-10-05

View original

CI License Bun Deploy to Cloudflare Deploy with Vercel — Vercel quickstart

Docker run → arra mine ~/notes → open the UI → search your memory.

Arra Oracle is the Oracle family's local memory and search layer. It stores your notes in SQLite, searches them with FTS/vector-capable APIs, and exposes the same memory through HTTP, MCP, the arra CLI, plugins, and the Studio UI.

Quick start: Docker is the primary path

You only need Docker, curl, and a notes folder. This path starts the HTTP server, mines ~/notes, opens the built-in UI, and performs the first search. No local Bun install, API key, schema choice, or vector service is required.

1. Run Arra Oracle

export ARRA_PORT="${ARRA_PORT:-47778}"
export ARRA_URL="http://127.0.0.1:${ARRA_PORT}"
export ARRA_CONTAINER="${ARRA_CONTAINER:-arra-oracle}"
export ARRA_VOLUME="${ARRA_VOLUME:-arra-oracle-data}"
export ARRA_NOTES_DIR="${ARRA_NOTES_DIR:-$HOME/notes}"

mkdir -p "$ARRA_NOTES_DIR"
docker volume create "$ARRA_VOLUME" >/dev/null

docker run --rm -d --name "$ARRA_CONTAINER" \
  -p "${ARRA_PORT}:47778" \
  -v "${ARRA_VOLUME}:/data" \
  -v "${ARRA_NOTES_DIR}:${ARRA_NOTES_DIR}:ro" \
  ghcr.io/soul-brews-studio/arra-oracle-v3:http

until curl -sf "${ARRA_URL}/api/health" >/dev/null; do sleep 1; done
echo "Arra Oracle is ready: ${ARRA_URL}"

If port 47778 is busy, run export ARRA_PORT=47878 first. If your notes live somewhere else, set ARRA_NOTES_DIR=/path/to/notes before docker run.

2. Mine your notes

Use the CLI bundled inside the running container so ingestion writes to the same Docker volume as the server:

arra() {
  docker exec "$ARRA_CONTAINER" bun dist-cli/index.js "$@"
}

arra mine ~/notes

Re-running arra mine is safe: unchanged Markdown, MDX, and text files are skipped with deterministic IDs. If you pointed ARRA_NOTES_DIR somewhere else, run arra mine "$ARRA_NOTES_DIR".

3. Open the UI

Open Simple Mode in your browser:

echo "${ARRA_URL}/simple"

Simple Mode shows health, save/search actions, and links to advanced surfaces.

Use the UI search box, or call the HTTP API directly:

curl -sfS "${ARRA_URL}/api/v1/search?q=runbook&mode=fts&limit=5"

For grounded answers with citations:

curl -sfS "${ARRA_URL}/api/v1/ask" \
  -H 'content-type: application/json' \
  -d '{"q":"What did I write about runbooks?","limit":5,"llm":false}'

"llm": false keeps the answer extractive and local.

Stop, restart, or inspect

curl -sf "${ARRA_URL}/api/health"
docker logs "$ARRA_CONTAINER"
docker stop "$ARRA_CONTAINER"

Restart later by re-running the docker run block. Your memory remains in the Docker volume named by $ARRA_VOLUME.

MCP clients with Docker

Use the stdio image when a desktop or agent needs Oracle MCP tools. It can share the same Docker volume as the HTTP server:

claude mcp add arra-oracle -- docker run --rm -i \
  -e ORACLE_LOG_TARGET=stderr \
  -v "${ARRA_VOLUME:-arra-oracle-data}:/data" \
  ghcr.io/soul-brews-studio/arra-oracle-v3:stdio

claude mcp list

The MCP surface includes Oracle search, read, learn, recap, profile, research note, trace, and tool-catalog capabilities.

What ships

Area What it gives you
Docker HTTP image Long-running local server on port 47778 with SQLite data in /data.
arra mine First ingestion path for folders of .md, .mdx, and .txt notes.
Simple Mode UI Browser entry point at /simple for health, save, and search.
HTTP API /api/v1/search, /api/v1/ask, /api/v1/learn, vector status, plugins, menu, and MCP tool discovery.
MCP server Stdio tool server for Claude, Codex, Docker MCP Toolkit, and agent fleets.
Memory contracts Confidence-ranked retrieval, reversible supersede history, provenance, and tenant-scoped reads/writes.
Plugins Unified manifests for CLI commands, API/menu rows, MCP tools, sidecars, exports, and lifecycle hooks.
Edge/frontends Cloudflare Worker shapes, Vercel Studio proxy, React/Tauri Studio, and canvas surfaces.

Architecture at a glance

Notes / agents / browsers / MCP clients
        │
        ├── Docker HTTP: ghcr.io/...:http on :47778
        ├── Docker stdio MCP: ghcr.io/...:stdio
        ├── CLI: arra mine/search/learn/export
        └── Studio and Simple Mode UI
                  │
        Elysia routes + MCP tools + plugin registry
                  │
        SQLite + FTS + optional vector stores + local vault files

The design goal is one memory core with thin adapters. CLI, HTTP, MCP, plugins, canvas, and web/desktop surfaces reuse shared contracts instead of duplicating business logic.

Source development path

Use a local checkout only when editing Arra Oracle itself:

git clone https://github.com/Soul-Brews-Studio/arra-oracle-v3.git
cd arra-oracle-v3
bun install
bunx tsc --noEmit
bun run server

Then open http://localhost:47778/simple or call http://localhost:47778/api/v1/search?q=oracle&mode=fts.

Useful checks before a PR:

bunx tsc --noEmit
bun test tests/http/<cluster>/
bun test tests/docs/link-checker.test.ts

Work targets alpha; never push or merge directly to main. Keep source, test, and docs files at or below 250 lines.

More docs

License

Arra Oracle V3 is licensed under BUSL-1.1. See LICENSE.

Acknowledgments

Inspired by claude-mem by Alex Newman — process manager patterns, worker service architecture, and hook system concepts.