Aligned Groups Framework — Basic

The Aligned Groups Framework (AGF) in its most portable form: the official controlled vocabulary, machine-readable extractions of it, and a single self-contained web page that introduces the framework's core.

The AGF is a structured ethical framework designed to articulate essential elements in the process of humans ethically aligning with each other, especially those requiring explicit articulation when integrating AI systems into organizational processes. — ~3-35

Aligned Groups Framework © 2024 by The Understanding Group, licensed under CC BY-SA 4.0. Use of the AGF is free but requires attribution, and is not a guarantee of alignment with the AGF.

---

What's here

Path	What it is
site/index.html	Start here. A one-page introduction to the AGF Core and the EGM pace layers. Open it in any browser — no server, no build, no internet connection.
vocabulary/	The official controlled vocabulary: the workbook plus seven TSV extractions.
docs/USING-THE-VOCABULARY.md	How the vocabulary is structured, how to cite it, and how to ground an AI assistant in it.
scripts/	Three small utilities that produced the published files, so you can reproduce them.
LICENSE	The full CC BY-SA 4.0 legal code, covering the framework and its content.
scripts/LICENSE	MIT, covering the three utility scripts only.
NOTICE	Attribution details and provenance.

The page

Open site/index.html by double-clicking it. That is the whole setup.

It is a single file with no dependencies of any kind — no fonts, scripts, stylesheets, images, analytics or trackers are loaded from anywhere. It works from file://, from a USB stick, from a corporate network with everything blocked, and offline. It prints cleanly, adapts to phone screens, respects your system's light/dark preference, and is keyboard navigable.

To publish it, copy site/ to any static host — GitHub Pages, S3, a shared drive. There is nothing to configure.

The page covers the framework's core only: the three imperatives and their intersections, the Ethics–Governance–Management pace layers, the integration gap, and the baseline / tension / flourishing distinction. It is deliberately not a tour of the whole framework.

The Venn diagram uses the same geometry and label engine as the canonical AGF website — a true equilateral three-circle layout, with each label centred on its region's scanned centre and auto-sized until the whole text block provably fits inside that region, so a curved petal boundary can never clip a word. The website computes this in the browser; because this page must work with JavaScript disabled, scripts/build_venn.py runs the same algorithm at build time and emits static SVG:

python scripts/build_venn.py --report   # the computed centre and font size per region
python scripts/build_venn.py            # the <svg> body, to paste back into site/index.html

Edit the labels there and regenerate rather than hand-tuning coordinates in the HTML.

The vocabulary

129 terms, each with a stable Unique Record ID in the form ~3-41 — where 3 identifies the AGF vocabulary and 41 the term. Cite by URID and your reference survives later refinements of the wording.

vocabulary/
├── Aligned Groups Framework Official Controlled Vocabulary v.1.0.xlsx   ← the workbook
└── tsv/
    ├── AGF_Choreography_Key.tsv        identity: which vocabulary, which term
    ├── AGF_Ontology_A.tsv              the basic definition of each term  ← start here
    ├── AGF_Ontology_B.tsv              examples of appropriate use and of misuse
    ├── AGF_Taxonomy.tsv                elaboration; broader/narrower/related terms
    ├── AGF_Choreography_Internal.tsv   editorial state
    ├── AGF_Choreography_External.tsv   source IDs, related images, editor's notes
    └── AGF_Bibliography.tsv            245 cited sources, keyed S~N

The workbook is the source of truth; every TSV is a mechanical projection of it, sliced along the vocabulary's own information architecture. Each slice repeats Unique Record ID (URID) and Term, so any two files join cleanly on the URID.

Full column schema, notation (including <<ghost>> terms), and AI-grounding guidance: docs/USING-THE-VOCABULARY.md.

Reproducing the extractions

The TSVs are generated, not hand-maintained. To regenerate or verify them:

pip install openpyxl
python scripts/export_tsv.py            # regenerate
python scripts/export_tsv.py --check    # verify only; exits 1 on drift

Provenance and canonical sources

The canonical working version of the vocabulary is a Google Sheet maintained with Apps Script; the workbook here is the Excel export of it, published for reference. The AGF's visual models are maintained on a public Miro board, linked from the workbook's About and Controlled Visual Models sheets.

The published workbook has had its editorial scaffolding removed — reviewer comment threads and the reviewer-identity part that backs them. No cell was changed: the strip is verified cell-by-cell against the source, and you can re-run that verification yourself:

python scripts/clean_workbook.py SOURCE.xlsx OUTPUT.xlsx --verify

Attribution

When you use or adapt the AGF, CC BY-SA 4.0 requires attribution and share-alike. A minimal attribution line:

Aligned Groups Framework © 2024 The Understanding Group, CC BY-SA 4.0. https://creativecommons.org/licenses/by-sa/4.0/

See NOTICE for the full attribution, the acknowledged collaborators, and the AI-use disclosure that accompanies the vocabulary.

License

Two licences, split along the content/code line:

What	Licence
The framework — the controlled vocabulary, the workbook, the TSVs, the documentation, and the introduction page	CC BY-SA 4.0 — see LICENSE
The three utility scripts in scripts/	MIT

The scripts are build tooling, not part of the framework, so they carry a permissive code licence: reuse the TSV exporter or the workbook cleaner in your own project without ShareAlike obligations. Anything you build from the vocabulary remains under CC BY-SA 4.0.