the_y_coding_standard

An opinionated coding-standards plugin for Claude Code. Hard rules for Python and React/Next.js. Drop-in AGENTS.md template. Parallel governance audit inspired by Microsoft's agent-governance-toolkit.

Install

# add the marketplace
/plugin marketplace add yashs33244/the_y_coding_standard

# install the plugin
/plugin install the-y-coding-standard

What you get

Surface	What it does
the-y-coding-standard skill	Aggressive auto-trigger on new code, new features, and standards-related prompts. Loads the right reference file per stack.
/y-init command	Scaffolds the agentic engineering stack into the current repo: AGENTS.md, CLAUDE.md, .claude/, .agents/, .conductor/, .cursor/, .devcontainer/, .github/, .husky/, Docs/, plans/, findings/, progress/.
/y-govern command	Dispatches three parallel sub-agents (security, audit, permissions) inspired by Microsoft's agent-governance-toolkit. Writes a findings report.

Reference Files

Loaded by the skill on demand:

• references/python.md: Python version policy (latest stable, looked up at runtime), uv + Poetry, conda envs, modularity rules (separate enums.py / constants.py / types.py), pydantic boundaries, error handling, 20+ common pitfalls.
• references/react-nextjs.md: Next.js App Router 14+, Server Components default, 400-line component cap, useReducer over useState explosion, centralized config, middleware patterns, Vercel AI SDK v6 defaults.
• references/agentic-stack.md: AGENTS.md format, the canonical folder tree, plans/findings/progress/ semantics, the Skillify 10-item promotion checklist.
• references/governance.md: The five governance dimensions, parallel sub-agent dispatch model, policy catalog, permission matrix template.
• references/project-scaffold.md: Every dotfolder Yash uses (.claude, .conductor, .cursor, .devcontainer, .github, .husky, Docs), what each one does, canonical config inside.

Quickstart

# in any repo
/y-init           # scaffolds the agentic stack
/y-govern         # runs the parallel governance audit

Standards Highlights

Python:

• Latest stable Python at scaffolding time (looked up via WebSearch).
• uv default, Poetry fallback, conda for env management.
• Each enums / constants / types / exceptions in its own file.
• Pydantic at API boundaries, dataclass internally.
• Error handling on every external call.
• ruff for lint + format. mypy --strict. pytest with 90% coverage gate.

React / Next.js:

• App Router 14+, never pages/.
• Server Components default, push client components to leaves.
• 400-line component hard cap.
• useReducer when state has 3+ related fields.
• Centralized config via lib/config.ts with zod schema.
• Middleware for cross-cutting concerns.
• Vercel AI SDK v6 by default for AI features.

Agentic Stack:

• AGENTS.md is the universal agent contract.
• plans/ for forward-looking docs, findings/ for investigations, progress/ for journals.
• Issue -> Plan -> Implement -> Test -> Review -> Ship -> Retro loop.

Governance:

• Run /y-govern before every release.
• Five dimensions: security, policy enforcement, auditability, compliance, agent permissions.
• Three parallel sub-agents (security, audit, permissions) inspired by Microsoft agent-governance-toolkit.
• Findings written to findings/YYYY-MM-DD-governance.md.

Compatibility

• Claude Code (current).
• Python 3.13+ required for /y-govern scripts (uses uv run for the inline metadata).
• Node 20+ recommended for Next.js scaffolding.

License

MIT.

Inspirations

• Anthropic Claude Code skill format
• Vercel Labs agent-skills (https://github.com/vercel-labs/agent-skills)
• Microsoft agent-governance-toolkit (https://github.com/microsoft/agent-governance-toolkit)
• Microsoft AI-Engineering-Coach (https://github.com/microsoft/AI-Engineering-Coach)
• Skillify checklist pattern (Stanford agentic-workflows talk)

Contributing

Open a PR. Run /y-govern first.

Author

https://github.com/yashs33244