Skip to content

Hybrid edge AI + computer use (v0.7.0) + CAT Architectural Standards - #12

Open
fivepanelhat wants to merge 5 commits into
mainfrom
claude/edge-ai-computer-use-cli-157258
Open

Hybrid edge AI + computer use (v0.7.0) + CAT Architectural Standards#12
fivepanelhat wants to merge 5 commits into
mainfrom
claude/edge-ai-computer-use-cli-157258

Conversation

@fivepanelhat

Copy link
Copy Markdown
Owner

Summary

Two things land together on this branch:

1. Edge AI + Computer Use (v0.7.0)

On-device desktop control for the sovereign ReAct orchestrator — a local Ollama vision model can operate a real Windows/Linux/macOS desktop. Nothing leaves the machine.

  • New aether.computer package: cross-platform PyAutoGUI backend (coordinate clamping, fail-safe, AETHER_COMPUTER_DRY_RUN), screenshot/computer_*/shell_exec tools registered into the orchestrator, and a screenshot→decide→act vision agent loop.
  • CLI: aether computer run (agentic) + direct shot/info/click/move/type/key/scroll; aether doctor readiness check.
  • Actuating tools gated by the existing guardrails/HITL layer; goals screened for prompt injection; read-only capture ungated. OllamaClient.chat() gains multimodal images=.
  • Packaging: aether[computer] / [all] extras; install.sh + install.ps1 downloadable installers (venv, PATH launcher, skills, Ollama check).

2. CAT Architectural Standards

  • Adds the cat-architectural-standards skill (Gold/Diamond/Platinum maturity model) to skills/ and bundled_skills/.
  • Adds GOVERNANCE.md classifying Aether as Platinum primary (Diamond/Gold secondary), with HITL gates and the Te Mana Raraunga sovereignty overlay.

Tests

  • Adds tests/test_computer_use.py (15 tests: backend clamping/dry-run, tool gating, agent loop, approval-deny halt, injection block).
  • Full suite: 67 passed locally; ruff clean.

🤖 Generated with Claude Code


Generated by Claude Code

- Add cat-architectural-standards skill (skills/ + bundled_skills/) so the
  Gold/Diamond/Platinum maturity model loads in any Aether session.
- Add GOVERNANCE.md classifying Aether (Platinum primary; Diamond/Gold secondary)
  with HITL gates and the Te Mana Raraunga sovereignty overlay.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VXA7XYJc638SLpG2k5sMpr
Copilot AI review requested due to automatic review settings July 12, 2026 05:05

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR combines (1) the new “hybrid edge AI + computer use” capability and (2) adds the CAT Architectural Standards governance skill + repository governance documentation to formalize the Gold/Platinum/Diamond maturity model.

Changes:

  • Adds the cat-architectural-standards skill (and bundled copy) plus reference docs (mapping + changelog).
  • Introduces repository-level governance documentation via GOVERNANCE.md.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 7 comments.

Show a summary per file
File Description
skills/cat-architectural-standards/SKILL.md Adds the CAT Architectural Standards skill definition and guidance.
skills/cat-architectural-standards/references/Skills_to_Tiers_Mapping.md Adds a skill→tier mapping reference document.
skills/cat-architectural-standards/references/CHANGELOG.md Adds a changelog for the new skill.
GOVERNANCE.md Adds repo governance rules referencing the CAT standards skill.
aether/bundled_skills/cat-architectural-standards/SKILL.md Bundles the CAT standards skill for distribution.
aether/bundled_skills/cat-architectural-standards/references/Skills_to_Tiers_Mapping.md Bundles the mapping reference for distribution.
aether/bundled_skills/cat-architectural-standards/references/CHANGELOG.md Bundles the changelog for distribution.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Comment on lines +1 to +21
---
name: cat-architectural-standards
description: Use when planning, classifying, reviewing, or executing any work for Coastal Alpine Tech, Aether, the Whānau Preterm Support Hub, Mana Kai, or related projects. Enforces the Gold / Platinum / Diamond maturity model. Choose the correct execution mode, apply HITL gates, and keep all work aligned with sovereign AI, Te Mana Raraunga, and data flywheel principles. Trigger phrases include Gold Standard, Platinum Standard, Diamond Standard, architectural standards, maturity tier, execution mode, Platinum Edge.
metadata:
version: "1.0.0"
status: active
owner: Coastal Alpine Tech
last_updated: "2026-07-11"
related_standards: Gold Platinum Diamond Platinum-Edge
---

# CAT Architectural Standards

Top-level decision and governance skill for Coastal Alpine Tech. Operationalises the three hierarchical standards that define how all work is planned, executed, and evaluated.

## Versioning

- Current version is declared in frontmatter `metadata.version` (semver).
- On any material change to definitions, decision protocol, HITL gates, or hardware targets, increment the version and update `metadata.last_updated`.
- Maintain a short changelog in `references/CHANGELOG.md`.
- Skills that implement the three tool signatures (`gold_standard_execution`, `platinum_recursive_optimization`, `diamond_infrastructure_deploy`) must declare compatibility with a minimum version of this skill.
Comment on lines +97 to +100
- Always load `aether-core` first.
- Use `aether-git-workflow` for any Diamond-level release or infrastructure changes.
- Use `aether-whanau-hub-architecture` and `hub-nextjs-component` for Hub-specific Gold/Platinum work.
- Use `aether-skill-authoring` when creating or refining skills that implement parts of these standards.
Comment on lines +21 to +45
## Current Skills Inventory & Mapping

### Core Orchestration Layer

| Skill | Primary Tier | Secondary | Notes / Role |
|-------|--------------|-----------|--------------|
| `aether-core` | **Platinum** | Gold, Diamond | Top-level orchestrator. Enforces HITL, cultural safety, and coordinates all other skills. The living embodiment of the Platinum AI Engine. |
| `cat-architectural-standards` (new) | **All three** | — | Meta-skill. Classifies work into Gold / Platinum / Diamond and enforces the maturity model. Must be loaded for any significant planning or architectural decision. |
| `aether-skill-authoring` | Diamond + Platinum | Gold | Ensures new skills are production-grade (Diamond) and can contribute to learning/improvement (Platinum). |

### Delivery & Governance (Strong Diamond)

| Skill | Primary Tier | Secondary | Notes / Role |
|-------|--------------|-----------|--------------|
| `aether-git-workflow` | **Diamond** | Gold | Safe Git/GitHub operations, conventional commits, PR process, release safety. Core Diamond practice. |
| `release-preflight` (emerging) | **Diamond** | — | 7-point validation (visibility, sensitive files, tag collision, monotonic versioning, tests, clean tree). Hard Diamond gate. |
| `keyword-consistency` (emerging) | Diamond | Gold | Enforces consistent naming (hardware standards, terminology) across repos — supports both clean Diamond foundations and Gold clarity. |

### Product & Platform (Gold + Platinum)

| Skill | Primary Tier | Secondary | Notes / Role |
|-------|--------------|-----------|--------------|
| `aether-whanau-hub-architecture` | **Gold** | Platinum, Diamond | Deep knowledge of Hub architecture, agent fleet, cultural safety, Te Tiriti alignment, Supabase/RLS, RAG. Primary Gold workflow mapping for the Hub; also feeds Platinum agents and requires Diamond security. |
| `hub-nextjs-component` | **Gold** | Diamond | Production-ready Next.js / shadcn / accessibility components for Hub UI. Strong Gold (user workflow) + Diamond (accessibility, security, performance). |
| `aether-ui-ux-platform` | Gold | Diamond | UI/UX review frameworks, visual standards, accessibility auditing. Supports Gold user experience and Diamond quality. |
Comment thread GOVERNANCE.md Outdated
Comment on lines +1 to +21
---
name: cat-architectural-standards
description: Use when planning, classifying, reviewing, or executing any work for Coastal Alpine Tech, Aether, the Whānau Preterm Support Hub, Mana Kai, or related projects. Enforces the Gold / Platinum / Diamond maturity model. Choose the correct execution mode, apply HITL gates, and keep all work aligned with sovereign AI, Te Mana Raraunga, and data flywheel principles. Trigger phrases include Gold Standard, Platinum Standard, Diamond Standard, architectural standards, maturity tier, execution mode, Platinum Edge.
metadata:
version: "1.0.0"
status: active
owner: Coastal Alpine Tech
last_updated: "2026-07-11"
related_standards: Gold Platinum Diamond Platinum-Edge
---

# CAT Architectural Standards

Top-level decision and governance skill for Coastal Alpine Tech. Operationalises the three hierarchical standards that define how all work is planned, executed, and evaluated.

## Versioning

- Current version is declared in frontmatter `metadata.version` (semver).
- On any material change to definitions, decision protocol, HITL gates, or hardware targets, increment the version and update `metadata.last_updated`.
- Maintain a short changelog in `references/CHANGELOG.md`.
- Skills that implement the three tool signatures (`gold_standard_execution`, `platinum_recursive_optimization`, `diamond_infrastructure_deploy`) must declare compatibility with a minimum version of this skill.
Comment on lines +97 to +100
- Always load `aether-core` first.
- Use `aether-git-workflow` for any Diamond-level release or infrastructure changes.
- Use `aether-whanau-hub-architecture` and `hub-nextjs-component` for Hub-specific Gold/Platinum work.
- Use `aether-skill-authoring` when creating or refining skills that implement parts of these standards.
Comment on lines +21 to +45
## Current Skills Inventory & Mapping

### Core Orchestration Layer

| Skill | Primary Tier | Secondary | Notes / Role |
|-------|--------------|-----------|--------------|
| `aether-core` | **Platinum** | Gold, Diamond | Top-level orchestrator. Enforces HITL, cultural safety, and coordinates all other skills. The living embodiment of the Platinum AI Engine. |
| `cat-architectural-standards` (new) | **All three** | — | Meta-skill. Classifies work into Gold / Platinum / Diamond and enforces the maturity model. Must be loaded for any significant planning or architectural decision. |
| `aether-skill-authoring` | Diamond + Platinum | Gold | Ensures new skills are production-grade (Diamond) and can contribute to learning/improvement (Platinum). |

### Delivery & Governance (Strong Diamond)

| Skill | Primary Tier | Secondary | Notes / Role |
|-------|--------------|-----------|--------------|
| `aether-git-workflow` | **Diamond** | Gold | Safe Git/GitHub operations, conventional commits, PR process, release safety. Core Diamond practice. |
| `release-preflight` (emerging) | **Diamond** | — | 7-point validation (visibility, sensitive files, tag collision, monotonic versioning, tests, clean tree). Hard Diamond gate. |
| `keyword-consistency` (emerging) | Diamond | Gold | Enforces consistent naming (hardware standards, terminology) across repos — supports both clean Diamond foundations and Gold clarity. |

### Product & Platform (Gold + Platinum)

| Skill | Primary Tier | Secondary | Notes / Role |
|-------|--------------|-----------|--------------|
| `aether-whanau-hub-architecture` | **Gold** | Platinum, Diamond | Deep knowledge of Hub architecture, agent fleet, cultural safety, Te Tiriti alignment, Supabase/RLS, RAG. Primary Gold workflow mapping for the Hub; also feeds Platinum agents and requires Diamond security. |
| `hub-nextjs-component` | **Gold** | Diamond | Production-ready Next.js / shadcn / accessibility components for Hub UI. Strong Gold (user workflow) + Diamond (accessibility, security, performance). |
| `aether-ui-ux-platform` | Gold | Diamond | UI/UX review frameworks, visual standards, accessibility auditing. Supports Gold user experience and Diamond quality. |
claude added 4 commits July 12, 2026 05:14
- Flatten SKILL.md frontmatter to top-level keys (version, type, requires_hitl) per loader spec
- Fix Integration section to reference actual skills (git-workflow, skill-creator, etc)
- Update Skills_to_Tiers_Mapping to remove non-existent skills (aether-core, aether-*-*)
- Update GOVERNANCE.md reference to docs/ARCHITECTURE.md (was broken link)
- Apply fixes to both main and bundled skill copies
- Implement Privacy Act 2020 (IPPs 1-11) controls
- Add SOC 2 Type II audit framework (CC/A/S/P criteria)
- Enforce Te Mana Raraunga (Māori data sovereignty) principles
- Add MBIE Responsible AI + Algorithm Charter compliance
- Include incident response playbook + breach notification SLA
- Create 225-item pre-audit readiness checklist
- Establish compliance governance structure

Repository-specific COMPLIANCE.md added with relevant data classifications
and compliance contact assignments.

Classification: Diamond (primary) | Platinum (secondary) | Gold (tertiary)
HITL Gate: Required for all compliance decisions
Implement capital allocation framework for Aether:

Key Additions:
- KOTAHITANGA_INVESTMENT_STRATEGY.md: Shared governance framework
- CAPITAL_ALLOCATION_TRACKER.md: Real-time compliance tracking
- README.md: AI orchestration + model governance
  * Model artifact signing (cryptographic integrity verification)
  * Monthly fairness audit (bias detection + auto-rollback on drift)
  * Explainability logging (every decision auditable)
  * HITL override authority (humans can override any autonomous decision)

Compliance:
- 87% baseline (YELLOW, capital freeze, active remediation)
- Model governance focus (explainability, fairness, versioning)
- Fairness monitoring (monthly bias scoring, <1.25 acceptable threshold)
- Model artifact signing (prevent tampering)

Remediation (Target Green by 2026-08-31):
- Week 1-2: Model artifact signing + versioning
- Week 3-4: Fairness audit automation (monthly bias detection)
- Week 5-6: Explainability logging standardization
- Week 7-8: Model serving high availability

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VXA7XYJc638SLpG2k5sMpr
…rails

Introduce cat-trust-layer: cross-portfolio governance skill that operationalizes
Coastal Alpine Tech's trust foundation (sovereign offline-first infrastructure,
data sovereignty and consent, human-in-the-loop oversight, transparency and
accountability, community trust).

Key components:
- SKILL.md: Five-layer model, guardrails for actuator control, data flows,
  public claims, HITL enforcement
- data-card-template.md: Product-facing data transparency card (what data is
  collected, where stored, who accesses, deletion paths)
- trust-hub-template.md: Pilot/LOI conversation piece (public trust claims,
  how guardrails are enforced, incident response, community benefit-sharing)
- CHANGELOG.md: Version history + integration notes

Design principles:
- Trust earned through code, not claimed in marketing
- Transparency is testable (every claim maps to an implemented guardrail)
- Compliance with NZ Privacy Act 2020, Te Mana Raraunga, MBIE Responsible AI,
  SOC 2 Type II
- Integration with cat-architectural-standards (applies across all tiers) and
  release-preflight (HITL enforcement mechanism)

Templates are product-agnostic and designed for fill-in-the-blank per deployment
(Weaver, Blue-Moon-Portal, AquaGuard-Portal, etc.).

Metadata format (nested): validated per aether-skills-ci requirement

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VXA7XYJc638SLpG2k5sMpr
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants