# SCALE Portable Agent Kit

**Strategic Capabilities Architecture Linkage Engine**  
**Version:** 0.5  
**Published:** August 2026  
**Delivery model:** Bring your own AI

## What this kit does

SCALE is a portable set of agent instructions, architecture rules, working registers, schemas, and examples. Load the kit into an AI environment approved by your organization, then use it to perform bounded, architect-led Business Architecture analysis.

The business architect remains the primary user and accountable interpreter. SCALE helps assemble evidence, propose architecture, trace relationships, expose gaps, assess confidence, and create work products. It does not establish enterprise truth or take formal decision rights.

## Before you begin

Use only an AI environment and account approved for the information you intend to provide. Do not upload employer, client, customer, employee, regulated, export-controlled, security-sensitive, or proprietary material to an unapproved service.

The kit can record access classifications supplied by you. It cannot enforce source-system permissions, retention rules, legal holds, or enterprise access controls inside a general AI product.

## Five-minute start

1. Unzip the kit into a folder you control.
2. Read `PLATFORM-SETUP.md` and choose the setup pattern supported by your AI.
3. Install or attach `SCALE-AGENT-INSTRUCTIONS.md` as the controlling instruction file.
4. Attach the files in `reference/` and `schemas/`. Attach the Northstar files in `examples/` only if you want a demonstration.
5. Open `LAUNCH-PROMPT.md`, replace the bracketed fields you already know, and send it to the AI.
6. Provide only the evidence needed for the first business question.
7. Review every proposed element, relationship, finding, and change. Use **Accept**, **Modify**, **Reject**, or **Investigate**.
8. Send `EXPORT WORKSPACE` before ending the session. Save the returned Markdown and JSON with your governed working files.

## Recommended loading order

1. `SCALE-AGENT-INSTRUCTIONS.md`
2. `reference/SCALE-REFERENCE-ARCHITECTURE.md`
3. `reference/SCALE-ANALYSIS-WORKBOOK.md`
4. `schemas/scale-workspace.schema.json`
5. `schemas/scale-workspace-starter.json`
6. Your authorized evidence
7. `LAUNCH-PROMPT.md`

## File map

| File or folder | Purpose |
|---|---|
| `SCALE-AGENT-INSTRUCTIONS.md` | Controlling behavior, workflow, review rules, commands, and output contract |
| `LAUNCH-PROMPT.md` | Copy-and-paste command that starts a governed analysis |
| `PLATFORM-SETUP.md` | Vendor-neutral installation patterns and capability checks |
| `WORKSPACE-EXPORT-TEMPLATE.md` | Human-readable continuity record for later sessions and governance |
| `reference/` | Version 0.5 analysis workbook and reference architecture |
| `schemas/` | JSON and CSV structures for portable architecture state |
| `examples/` | Fictional Northstar input case and completed SCALE walkthrough |
| `VERSION-HISTORY.md` | Product evolution and current release boundary |
| `SHA256SUMS.txt` | Integrity hashes generated when the downloadable package is built |

## Core commands

- `START ANALYSIS`
- `ADD EVIDENCE`
- `SHOW WORKSPACE`
- `REVIEW PROPOSALS`
- `TRACE [question or element]`
- `CREATE [work product]`
- `EXPORT WORKSPACE`
- `CONTINUE ANALYSIS`
- `CLOSE ANALYSIS`

Natural-language requests also work. The commands make state and intent explicit when the conversation becomes long.

## Definition of a successful first use

A first use is successful when the architect can inspect the evidence behind every material finding, distinguish fact from inference, record human disposition, produce a useful work product, and save accepted architecture for the next question.