DoDAF for Humans and AI

This repository is a vanilla upstream reference for using DoDAF 2.0 ideas to help humans and AI agents understand, explain, analyze, improve, and present systems.

The goal is not to turn every project into a paperwork factory. The goal is to give people and AI models a shared architecture language for questions like:

• What is this system trying to accomplish?
• What capabilities does it need?
• Who or what performs the work?
• What information moves, where, and under what rules?
• Which services, systems, standards, projects, timelines, and risks matter?
• What should humans decide, what can AI explain, and what still needs evidence?

This is not an official DoD repository and is not endorsed by the Department of Defense. The official source is the DoD CIO DoDAF 2.0 site.

License

This repository uses CC0 1.0 Universal to make the repository-created material as open as possible. To the extent any contributor holds copyright or related rights in the original docs, templates, prompts, schemas, scripts, or generated reference material in this repository, those rights are dedicated to the public domain under CC0-1.0.

The captured DoD CIO source material is included as public reference material from an official U.S. Government site. Under 17 U.S.C. § 105, copyright protection is generally not available for works of the United States Government. Separate rights may still apply to third-party, contractor, trademark, patent, privacy, publicity, or non-text media content, so treat official source notices and the live DoD CIO site as authoritative.

Do not imply Department of Defense, DoD CIO, or U.S. Government endorsement of this repository or downstream uses.

Hard Boundary

This repo must stay generic.

Do not store real project, customer, mission, environment, runtime, credential, contract, private, sensitive, or classified data here. Downstream architecture packages belong in the local project repositories that own that truth.

Use this repo for:

• Official-source snapshots and provenance.
• Reusable DoDAF model reference material.
• Blank templates.
• AI prompts, validation patterns, scripts, schemas, and automation ideas.
• Fictional examples only, if examples are added later.

Do not use this repo for:

• A specific system's actual architecture package.
• Real operational data, logs, diagrams, names, environments, endpoints, credentials, or evidence.
• Project-specific decisions or review notes.

What Is Here

• sources/dodcio-dodaf20/ - local browser-captured copy of the DoD CIO DoDAF 2.0 pages used as source material.
• reference/dodaf20/ - generated human-friendly reference catalog for all captured DoDAF-described models.
• templates/architecture-package/ - blank downstream architecture package template with one file per DoDAF-described model.
• docs/ - repo structure, AI/human method, and downstream adoption guidance.
• prompts/ - reusable AI prompt starters for model selection, consistency review, and presentation synthesis.
• schemas/ - generic schema foundations for future validation tools.
• tools/ - repeatable scripts for refreshing the official-source snapshot and rebuilding reference/templates.

Current snapshot:

• 60 official-source pages captured.
• 52 DoDAF-described model catalog rows captured.
• 52 blank model templates generated.

How To Use This

For a real project, copy templates/architecture-package/ into that project's repo and fill it in there. Keep this upstream repo clean and reusable.

Recommended downstream flow:

1. Start with the human decision, risk, explanation, or improvement need.
2. Fill AV-1 for scope and AV-2 for vocabulary.
3. Use MODEL_SELECTION.md to choose the smallest useful model set.
4. Populate only the selected artifacts in the project repo.
5. Use AI to summarize, trace gaps, check consistency, produce diagrams, and draft presentation views.
6. Keep human owners responsible for decisions, assumptions, approvals, and final truth.

Refreshing The Source Snapshot

The DoD CIO site may block plain local HTTP clients, so the snapshot tool uses the bundled Playwright browser CLI.

python3 tools/snapshot_dodcio_dodaf20.py --max-pages 90 --delay 0.2 --headed
python3 tools/build_dodaf20_reference.py

The first command captures raw HTML plus extracted HTML, Markdown, and text. The second command rebuilds the reference catalog and blank model templates from the captured model list.

Architecture Position

DoDAF is strongest when treated as a disciplined query system, not a compliance costume.

The official DoDAF model list is comprehensive but not prescriptive: decision-makers and process owners determine which models are required for their purposes. This repo follows that principle. It preserves the whole model universe so nothing gets lost, but downstream projects should select the artifacts that answer real questions.

AI can help by:

• Translating architecture artifacts into plain language.
• Asking better scoping and evidence questions.
• Finding inconsistency across viewpoints.
• Suggesting diagrams, matrices, reports, and presentations.
• Drafting skill/tool/automation patterns from architecture data.
• Identifying improvement opportunities and unresolved assumptions.

AI must not quietly invent architecture truth. If evidence is missing, it should say so.

Repo Status

The public repository is live on main as a maintainer-led reference repo.

• License recognized by GitHub: CC0-1.0.
• Public feedback path: GitHub Issues.
• Pull request creation: restricted to approved collaborators.
• Wiki and Projects: disabled.
• Branch protection: pull requests required for main, 0 approvals required for lone-maintainer flow, conversation resolution required, admin enforcement enabled, force pushes blocked, and branch deletion blocked.

Use issues for public suggestions and corrections. Use collaborator pull requests for repository changes.