BYOAI: Bring Your Own AI

The BYOAI Philosophy

IUENNA separates the research data from the AI model. The project provides authoritative repository metadata, persistent identifiers, explicit relationships, and open machine-readable interfaces. Researchers can then use the AI environment of their choice without depending on a single commercial platform.

ARCHE remains the authoritative long-term repository. The IUENNA web layer is a discovery and retrieval interface derived from ARCHE metadata and is designed to make archaeological research data easier to query reproducibly.

1. Model Context Protocol (MCP) v1.2

The IUENNA MCP server implements JSON-RPC 2.0 over stdio and runs with zero third-party runtime dependencies. Version 1.2.0 exposes seven research tools and now integrates semantic knowledge graph traversal alongside entity discovery and exhaustive primary-resource retrieval.

Setup for Claude Desktop / Cursor / compatible MCP clients

{
  "mcpServers": {
    "iuenna": {
      "command": "python3",
      "args": [
        "-c",
        "import urllib.request; exec(urllib.request.urlopen('https://raw.githubusercontent.com/IUENNA/IUENNA.github.io/main/mcp/server.py').read().decode())"
      ]
    }
  }
}

Alternative: Clone the repository and run python3 mcp/server.py or node mcp/index.mjs.

Available MCP Tools

2. Recommended AI Routing

3. Machine-Readable JSON & OpenAPI

IUENNA publishes static JSON snapshots over HTTPS. They require no project-specific API key. ARCHE remains authoritative for current repository metadata, access conditions, and rights information.

View openapi.json

4. llms.txt for LLM & Agent Discovery

The canonical llms.txt acts as a routing and provenance layer. It tells AI systems which endpoint to use for which type of question, establishes ARCHE as the authoritative repository, and specifies PID-first citation and rights handling.

https://iuenna.github.io/llms.txt

For exhaustive prompts such as “all files”, “all documentation”, or “all data concerning X”, an agent is explicitly instructed not to stop at the curated dataset catalogue or semantic discovery index, but to use arche_corpus.json and the collection hierarchy as well.

5. Minimal Research Recipes

Python: query places

import json
import urllib.request

url = "https://iuenna.github.io/data/arche_places.json"
with urllib.request.urlopen(url) as response:
    places = list(json.loads(response.read().decode("utf-8")).values())

hemmaberg = [p for p in places if "hemmaberg" in p.get("title", "").lower()]
print(json.dumps(hemmaberg, ensure_ascii=False, indent=2))

Python: explore semantic knowledge graph

import json
import urllib.request

url = "https://iuenna.github.io/data/arche_graph.json"
with urllib.request.urlopen(url) as response:
    graph = json.loads(response.read().decode("utf-8"))

nodes = {n["data"]["id"]: n["data"] for n in graph["elements"]["nodes"]}
edges = graph["elements"]["edges"]

# Find all publications authored by Franz Glaser
glaser_id = next((nid for nid, d in nodes.items() if "glaser" in d.get("label", "").lower()), None)
authored = [e["data"]["target"] for e in edges if e["data"]["source"] == glaser_id and e["data"]["predicate"] == "hasAuthor"]

for pub_id in authored:
    pub = nodes.get(pub_id, {})
    print(f"- {pub.get('label')} (PID: {pub.get('pid')})")

Python: exhaustive file-level search

import json
import urllib.request

url = "https://iuenna.github.io/data/arche_corpus.json"
with urllib.request.urlopen(url) as response:
    corpus = json.loads(response.read().decode("utf-8"))["resources"]

hits = [r for r in corpus if "georadar" in (
    f"{r.get('title','')} {r.get('filename','')} {r.get('description','')}".lower()
)]
print(json.dumps(hits[:20], ensure_ascii=False, indent=2))

MCP: relational query

{
  "name": "get_related_resources",
  "arguments": {
    "name_or_id": "Hemmaberg",
    "query": "Georadar",
    "limit": 50
  }
}

6. Provenance, Citation & Rights

ARCHE is the authoritative long-term repository. The top-level IUENNA collection is persistently identified by 21.11115/0000-0016-7B39-F.

Recommended project-level citation:

Hagmann, D., & Waldhart, F. (Eds.). (2025). IUENNA – openIng the soUthErn jauNtal as a micro-regioN for future Archaeology. ARCHE – Austrian Research Culture Heritage Extended. https://hdl.handle.net/21.11115/0000-0016-7B39-F

When using a specific dataset, document, image, GeoPackage, or other archived resource, cite its individual ARCHE PID and metadata rather than only the project homepage.