business_context - lgcorzo/rust_CACD_autonomous_factory GitHub Wiki


iso_doc_type: "Description" iso_viewpoint: "ContextView" type: "architecture" title: "Business Context — Dark Gravity CA/CD Autonomous Factory" description: "ISO 42010 ContextView / ISO 15289 Description documentation for Business Context — Dark Gravity CA/CD Autonomous Factory." tags: ['iso42010', 'okf', 'context_view', 'business'] timestamp: "2026-10-01T14:00:00Z" generated: "agent:okf-professional-documenter" verified: "true" last_verified_commit: "fbdc5848"

Business Context — Dark Gravity CA/CD Autonomous Factory

Purpose: Define the business rationale, innovation objectives, and value proposition of the Dark Gravity Autonomous Factory for stakeholders, investors, and compliance bodies.


1. Vision: Continuous Agentic / Continuous Deployment (CA/CD)

The Dark Gravity Autonomous Factory implements a paradigm shift from traditional CI/CD to CA/CD — Continuous Agentic / Continuous Deployment. In this model, AI agents autonomously:

  1. Ingest issues and PR comments from GitHub, GitLab, and Jira
  2. Plan implementation via Spec-Driven Development (SDD) with formal specifications
  3. Code solutions inside gVisor-sandboxed Kubernetes pods
  4. Validate changes through TDD (Red-Green-Refactor) and SAST security scanning
  5. Review via heuristic analysis and LLM-powered code judges
  6. Deliver code changes via PRs/MRs — never auto-merged (Human-in-the-Loop)

This transforms the software engineering lifecycle from a human-intensive, serial process into an agent-accelerated, parallel pipeline with formal governance.


2. Hazitek 2026 Innovation Objectives

The project is funded under the Hazitek 2026 program administered by SPRI (Sociedad para la Promoción y Reconversión Industrial) of the Basque Country government:

Objective Target Metric
Autonomous Code Generation ≥ 70% of routine tasks handled by agents Tasks completed / Total tasks
Zero-Toil Operations < 5 min human intervention per mission Avg. human time per mission cycle
Code Quality Assurance SAST score ≥ 8.0/10.0 on all agent output SastScanResult.score
Security Compliance 100% Ed25519 NHI credential coverage Verified credentials / Total missions
Financial Control < €50/day autonomous spend DailyBudgetConfig.max_daily_budget_usd
Documentation Currency OSR < 5% across all crates OsrMetric.osr_value

3. Return on Investment (ROI) Model

Cost Structure

Cost Center Description Tracking Mechanism
LLM Inference Token consumption across all agents FinOpsTag virtual headers + spend velocity alerts
Compute gVisor sandbox pod CPU/RAM usage Kubernetes resource quotas + core-hours telemetry
Infrastructure Kafka, Ziti, Vault, Hatchet hosting FluxCD GitOps cost attribution
Human Oversight HITL governance time at 4 vertices Sprint ceremony tracking

Value Drivers

Value Driver Quantification
Developer Time Savings Autonomous handling of lint fixes, test failures, and compilation errors
Time-to-Fix Reduction From hours (manual triage + fix + review) to minutes (agent detection → fix PR)
Quality Improvement Enforced TDD + SAST gates eliminate classes of defects before human review
Compliance Automation Automated R&D audit trail packaging replaces manual engineering timesheets
24/7 Operations Agents operate continuously; no on-call fatigue for routine remediations

4. KPI Matrix

mindmap
  root((Dark Gravity KPIs))
    Velocity
      Mission Throughput
      Time-to-Fix p50/p95
      Tasks per Day
    Quality
      SAST Score >= 8.0
      OSR < 5%
      Test Coverage
    Cost
      Daily Budget < 50 USD
      Token Efficiency
      Compute Core-Hours
    Governance
      HITL Compliance Rate
      Agent-Stuck Resolution Time
      PR Merge Latency

5. Stakeholder Map

Stakeholder Role Primary Interest
Product Owner HITL Vertex 1: Strategic Injection Business alignment, ROI, mission prioritization
Tech Lead HITL Vertex 2: Sprint Mobilization Task decomposition quality, sprint velocity
Architect HITL Vertex 3: Exception Override System integrity, deadlock resolution, architectural decisions
Senior Reviewer HITL Vertex 4: Categorical Scrutiny Code quality, security posture, merge approval
DevOps Engineer Operator Deployment stability, cluster health, troubleshooting
Security Auditor Compliance Zero Trust enforcement, NHI verification, SAST coverage
R&D Grant Evaluator Hazitek / SPRI Innovation output, compute investment, compliance evidence

6. Domain Vocabulary

See the comprehensive Glossary for all domain terms, architectural concepts, and technical acronyms used throughout this documentation.


Related: Strategic Design · HITL Governance · Compliance & Audit