AI-SDR (Software Development Runtime)

Turkish version: README_TR.md
Direct URL: https://github.com/boraeresici/AI-SDR/blob/main/README_TR.md

AI-SDR is a skill-driven runtime that keeps vibe-coding speed while enforcing SDD (Software Design Document) discipline end-to-end.

Simulation Status

• E2E process simulation completed on 2026-02-17.
• Process orchestration validation: PASS.
• Production release decision in simulation run: NO-GO (runtime checks pending).

Project Goal

• Control phase transitions: Blueprint -> Build -> Hardening
• Lock architecture, scope, UX, quality, operations, and ROI decisions with gates
• Standardize "fast delivery + low debt + measurable output"
• Prefer modular monolith with planned microservice extraction paths, backed by scaling and cost guardrails
• Prioritize performance-aware library selection and buy-over-build for commodity modules (auth, notifications, payments), with controlled dependency governance

Skill Ecosystem

• skills/loop-master-orchestrator/SKILL.md
• skills/idea-alchemist-market-strategist/SKILL.md
• skills/reverse-engineering-reference-architect/SKILL.md
• skills/architect-decision-matrix/SKILL.md
• skills/pm-context-scope/SKILL.md
• skills/ux-design-system/SKILL.md
• skills/fullstack-implementation-refactor/SKILL.md
• skills/qa-quality-gates/SKILL.md
• skills/devops-deployment-stability/SKILL.md
• skills/roi-investor-check/SKILL.md
• skills/contract-compat-check/SKILL.md
• skills/e2e-execution-gate/SKILL.md
• skills/security-review-gate/SKILL.md
• skills/architecture-drift-review/SKILL.md

Canonical Standards

• Skill format standard: skill-spec/SKILL_SPEC_v1.0.md
• Refactor plan: skill-spec/SKILL_REFACTOR_PLAN_v1.md
• Skill improvement backlog: skill-spec/SKILL_IMPROVEMENT_BACKLOG.md
• Multi-provider skill manifests: skills/*/agents/{manifest,openai,claude,gemini}.yaml
• Process diagram (render): process.md
• Process source: process.mmd
• User guide: USERGUIDE.md
• Standalone skill usage: USERGUIDE.md (Standalone Skill Calistirma)
• Artifact output standard: artifacts/YYYY-MM-DD/<phase>/
• Local validation: bash scripts/validate-skills.sh

Repository Structure

.
├── README.md
├── README_TR.md
├── .github
│   └── workflows/
│       └── validate-skills.yml
├── process.md
├── process.mmd
├── scripts
│   └── validate-skills.sh
├── artifacts
│   └── YYYY-MM-DD
│       ├── research/
│       ├── blueprint/
│       ├── build/
│       ├── hardening/
│       ├── release/
│       └── post-release/
├── domains
│   ├── INDEX.md
│   ├── edtech.md
│   ├── fintech-defi.md
│   ├── healthtech-senior-care.md
│   ├── martech-customer-analytics.md
│   ├── saas-productivity.md
│   ├── senior-care.md
│   └── smb-booking.md
├── skill-spec
│   ├── SKILL_SPEC_v1.0.md
│   ├── SKILL_REFACTOR_PLAN_v1.md
│   └── SKILL_IMPROVEMENT_BACKLOG.md
├── templates
│   ├── decision-matrix.md
│   ├── scope-pack.md
│   └── quality-gate-report.md
└── skills
    ├── <skill-name>/
    │   ├── SKILL.md
    │   └── agents/
    │       ├── manifest.yaml
    │       ├── openai.yaml
    │       ├── claude.yaml
    │       └── gemini.yaml
    └── ...

Process Summary

1. Idea Maturation:
   • Idea Alchemist & Market Strategist turns raw ideas into strategy-ready inputs.
2. Research Loop (Optional):
   • Reverse Engineering & Reference Architect deconstructs reference repos/docs into adaptation-ready artifacts.
3. Blueprint Loop:
   • Loop Master runs Domain Constraint Pre-check using domains/INDEX.md + selected domain doc.
   • If domain redlines fail, process is BLOCKED and idea/scope must be refined.
4. Build Loop:
   • Designer + Developer execute under quality and architecture constraints.
5. Hardening Loop:
   • QA + Contract + E2E + Security + DevOps gates run before release.
6. ROI + Drift:
   • ROI decision and post-release architecture drift checks.

Architecture Principles

• Choose FE/BE frameworks by measurable performance and operational cost, not trend bias.
• For commodity capabilities, prefer proven libraries/services before custom implementation.
• Every new dependency must include security, maintenance, performance, and lock-in impact notes.

Process Diagram

• Rendered file: process.md
• Open in GitHub: process.md

Artifact Policy

• Skill definitions are reusable, so project-specific outputs are never written under skills/.
• All phase outputs are written under artifacts/YYYY-MM-DD/<phase>/.
• Loop Master validates required artifact presence on each phase transition.

Project Docs

• Contribution guide: CONTRIBUTING.md
• Code of conduct: CODE_OF_CONDUCT.md
• Security policy: SECURITY.md
• Changelog: CHANGELOG.md
• Skill validation workflow: .github/workflows/validate-skills.yml

Templates

• templates/decision-matrix.md
• templates/scope-pack.md
• templates/quality-gate-report.md

Domains

• domains/INDEX.md (active domain docs + versions)
• domains/senior-care.md
• domains/smb-booking.md
• domains/healthtech-senior-care.md
• domains/fintech-defi.md
• domains/saas-productivity.md
• domains/martech-customer-analytics.md
• domains/edtech.md

Quick Start in 5 Minutes

1. Start Blueprint:
   • Loop Master, start Blueprint Loop in strict mode for [idea]. Domain: [domains/*.md]. Goal: [...]. Scope: [...]. Success metric: [...].
2. Run architecture decision:
   • Architect, prepare a 2-option MVP decision matrix and initial API/Auth/Data contracts.
3. Lock scope:
   • PM, produce scope-pack with MoSCoW and keep Must <= 60%.
4. Produce UI constraints:
   • Designer, generate ui-constraints with WCAG 2.1 AA and 375/768/1024/1440 breakpoints.
5. Continue flow:
   • Developer -> QA -> Contract -> E2E -> Security -> DevOps -> ROI -> Drift

License

This repository is licensed under MIT. See LICENSE.