The IRGF Practical Guide
The part of the documentation you use with a system in front of you: sector overlays, fill-in templates, working guidelines and checklists you can print.
Chapters 1 to 37 explain the framework: what each gate decides, how the dimensions are scored, why the artifact set is the size it is. This part assumes you have read enough of that to start, and answers the next question — what do I actually write down, in my sector, this week.
Nothing here overrides the chapters. Where the Practical Guide and a chapter appear to disagree, the chapter governs and the discrepancy is a defect worth reporting. Everything in this part is advice at the level the framework marks [Practice recommendation]: useful defaults, not requirements of IRGF.
The four sets
Industry overlays →
Ten sectors, each with tier calibration, the regulatory interface, gate additions, runtime signals, the failure modes that sector produces, and a ninety-day start.
Templates →
Every primary record as a form: purpose, owner, trigger, mandatory fields by tier, retention, related records, a filled example and a copyable plain-text version.
Guidelines →
Six working guidelines on the judgement calls the framework cannot make for you, each with the failure it exists to prevent.
Checklists →
Seven printable checklists covering the gates and the recurring operational work, with progress tracking that survives a page reload.
If this is your first week
The order below produces something defensible fastest. It assumes one architect and one risk lead, part-time, with no tooling.
- Read Chapter 27, the minimum viable framework, and decide what you are not doing this year.
- Open your sector's overlay and read section 1 and section 3. Twenty minutes.
- Inventory what exists. Use the classification checklist and expect the inventory to be the hardest part.
- Score three systems with the Risk Classification Record, with an independent countersigner in the room.
- Run one G3 dry run on a system already in production. The gap list is your real backlog.
- Wire one runtime signal — the one your overlay names first. Not a platform, one signal.
Status caveat. Everything in this part inherits the status of the framework itself: IRGF has not been deployed in a real organization. The sector overlays are reasoned from published regulation and common practice, not from field calibration, and the tier examples in them are illustrations of how the anchors read — not classifications you can adopt. Score your own systems. If a row in an overlay is wrong for your organization, that is exactly the feedback the project needs.
Conventions in this part
| Convention | Meaning |
|---|---|
| Tier badges in diagrams | The tier systems of that kind typically reach in that sector. Not a shortcut around scoring, and not a floor. |
| Illustrative D1–D5 rows | How the anchors in Chapter 6 read against sector facts. A range means it depends on a design choice named in the same row. |
| “The tell” and “the response” | Failure modes are written as a detectable symptom and the action that answers it, because a failure mode you cannot detect is not actionable. |
| Ticks in checklists | Stored in your browser only, per checklist. Nothing is transmitted; clearing site data clears them. |
| Plain-text template blocks | Copyable versions for pasting into a wiki, a form builder or an issue template. The HTML page is not the record. |