Deterministic AI context scaffolding for modern development environments.
Define the contract. Govern the AI.
AI-powered editors are powerful — but they're inconsistent without clear constraints. DevContextForge generates structured, role-driven AI development context for your repository so your AI tools follow a defined contract instead of guessing.
Think of it as ESLint for AI behavior — standardized, versionable, deterministic.
npx devcontextforge initor with flags:
npx devcontextforge init --env cursor --role senior-developer --strict ultra| Environment | Generated Files |
|---|---|
| Cursor | .cursorrules, .cursor/rules/devcontextforge.mdc, skills.md |
| VS Code / GitHub Copilot | .github/copilot-instructions.md, .github/skills/<role>/SKILL.md, .vscode/settings.json, .vscode/extensions.json, skills.md |
| Claude Code | CLAUDE.md, .claude/rules/code-style.md, .claude/rules/security.md, .claude/rules/testing.md, skills.md |
| Cloud IDE | .devcontainer/devcontainer.json, skills.md |
| Role | Level | Focus |
|---|---|---|
| 🌱 Junior Developer | 0–2 yrs | Learning fundamentals, mentorship-oriented AI |
| 🚀 Mid-Level Developer | 2–5 yrs | Independent feature ownership |
| ⚡ Senior Developer | 5–8 yrs | Production-grade, explicit error handling |
| 🏗️ Staff Engineer | 8–12 yrs | Cross-team impact, platform engineering |
| 🎯 Principal Engineer | 12+ yrs | Org-wide strategy, Conway's Law awareness |
| Role | Focus |
|---|---|
| 🎨 Frontend Engineer | Accessibility (WCAG 2.1), Core Web Vitals, component-driven |
| ⚙️ Backend Engineer | APIs, reliability, distributed systems, observability |
| 🔄 Full Stack Engineer | End-to-end ownership, shared TypeScript types |
| 🛠️ DevOps / Platform Engineer | IaC, GitOps, SRE, developer experience |
| 📊 Data / ML Engineer | Data quality, ML lifecycle, pipeline engineering |
| Role | Focus |
|---|---|
| 🧭 Tech Lead | Hands-on squad leadership, team standards |
| 🏛️ Software Architect | ADRs, formal design reviews, cross-cutting concerns |
| 👥 Engineering Manager | People, delivery, team health |
| 🔒 Security / AppSec Engineer | Threat modeling, OWASP, shift-left security |
| ✏️ Custom Role | Define your own with a free-text description |
| Level | Behavior |
|---|---|
| ⚖️ Balanced | AI has creative freedom within role guidelines. Best for exploration. |
| 📏 Strict | AI must justify deviations and follow standards rigorously. |
| 🔒 Ultra | Maximum enforcement. Every output must be production-ready. Best for critical systems. |
skills.md # Universal AI contract
.cursorrules # Legacy Cursor rules (widely supported)
.cursor/
rules/
devcontextforge.mdc # New .mdc format with YAML frontmatter
skills.md
.github/
copilot-instructions.md # Repository-level Copilot context
skills/
senior-developer/
SKILL.md # Agent Skill (agentskills.io spec)
.vscode/
settings.json # Workspace settings with Copilot config
extensions.json # Recommended extensions
skills.md
CLAUDE.md # Main project memory file
.claude/
rules/
code-style.md # Path-scoped: src/**/*
security.md # Path-scoped: api/**/*
testing.md # Path-scoped: **/*.test.*
- Run
devcontextforge init— interactive prompts guide you through environment, role, and strictness selection - Role profile is loaded — one of 14 built-in profiles (or your custom description)
- Strictness modifier is applied — adds progressive AI contract rules
- Templates are rendered — environment-specific files generated using your role + config
- Files are written — commit to version control; AI tools apply the contract on every run
devcontextforge init [options]
Options:
--env <environment> cursor | vscode | claude-code | cloud
--role <role> junior-developer | mid-developer | senior-developer |
staff-engineer | principal-engineer | frontend-engineer |
backend-engineer | fullstack-engineer | devops-engineer |
data-engineer | tech-lead | architect |
engineering-manager | security-analyst | custom
--strict <level> balanced | strict | ultra (default: balanced)
--dry-run Preview files without writing
--force Overwrite existing files
--cwd <path> Target directory (default: current directory)
-v, --version Show version
-h, --help Show help
# AI Development Context
# Role: ⚡ Senior Developer | Strictness: strict
## AI Operating Rules
- No speculative abstractions — implement what is needed today
- Explain architectural tradeoffs when multiple approaches exist
- Do not refactor without explicit justification of the benefit
- Flag any code that bypasses existing abstractions or patterns
- Require justification before refactoring existing code [strict]
- Enforce type annotations on all function signatures [strict]- Fork the repository
- Create a feature branch:
git checkout -b feat/my-role - Add your role JSON to
roles/following the schema - Submit a PR with clear description and tests
Contribution areas:
- New role profiles
- Additional IDE environment adapters
- Template improvements
- CLI enhancements
MIT © DevContextForge Contributors
DevContextForge — AI tools should follow a contract, not guess.