Context management, project guidelines, and code review for AI coding assistants
npx claudepluginhub xanderscannell/claude-cortexContext management, project guidelines, and code review for AI coding assistants. Prevent context degradation, enforce coding standards, and get structured code reviews — all in one plugin.
Context management, project guidelines, and code review for AI coding assistants — all in one plugin.
Three systems that work independently, but integrate deeply when used together:
Compatible with Claude Code and GitHub Copilot.
| Guide | What it covers |
|---|---|
| Context Framework | What CDS is, how .context/ works, daily workflow, templates |
| Project Guidelines | Simple vs full mode, severity system, review heuristics, subagent |
| Code Review | /code-review vs /ult-code-review, severity tiers, 10-agent suite, testability assessment |
# Marketplace
/plugin marketplace add xanderscannell/claude-cortex
/plugin install claude-cortex
# Or manual
git clone https://github.com/xanderscannell/claude-cortex.git
/plugin marketplace add ./claude-cortex
/plugin install claude-cortex
# Open the Command Palette (F1) and run:
Chat: Install Plugin From Source
# then point to this directory.
/cortex-init # Full setup: .context/ framework + PROJECT_GUIDELINES.md
# Or run them separately:
/context-init # Create .context/ framework + bootloaders
/guidelines-init # Generate PROJECT_GUIDELINES.md
/context-session # Load project context at session start; update at session end
/code-review # Review current changes
/guidelines-review --pr # Check PR against project guidelines
| Command | What it does |
|---|---|
/cortex-init | Full framework setup — runs /context-init then /guidelines-init in series, with a single combined commit |
/context-init | Scaffold .context/ directory, CLAUDE.md, and .github/copilot-instructions.md with auto-detected project details |
/context-session | Load context at session start, update status and guidelines at session end |
/guidelines-init | Generate PROJECT_GUIDELINES.md — simple (flat rules) or --full (categorized + severity) |
/guidelines-review | Review code against guidelines. Scopes: --pr, --staged, --changed, file, or directory |
/code-review | Security-first code review with 4 severity tiers, testability analysis, and merge verdict |
/ult-code-review | 10 parallel specialist agents for near-complete bug and vulnerability coverage |
claude-cortex/
|-- .claude-plugin/
| |-- plugin.json # Plugin manifest
|-- skills/
| |-- cortex-init/ # /cortex-init — full framework setup
| |-- context-init/ # /context-init — scaffold context framework
| |-- context-session/ # /context-session — load/update session context
| |-- guidelines-init/ # /guidelines-init — generate guidelines
| |-- guidelines-review/ # /guidelines-review — compliance review
| |-- code-review/ # /code-review — structured code review
| |-- ult-code-review/ # /ult-code-review — 10-agent parallel review
|-- agents/
| |-- reviewer.md # Subagent for parallel guideline reviews (51+ files)
|-- templates/
| |-- context/ # Templates for .context/ files
| |-- CLAUDE.md # Bootloader template (Claude Code)
| |-- copilot-instructions.md # Bootloader template (GitHub Copilot)
| |-- CURRENT_STATUS.md # Project status template
| |-- ARCHITECTURE.md # System design template
| |-- CONVENTIONS.md # Tooling and environment setup template
| |-- DECISIONS.md # ADR template
| |-- MASTER_PLAN.md # Roadmap template
| |-- SETUP.md # Dev setup template
|-- references/
| |-- guideline-categories.md # Per-stack example rules (JS/TS, Python, Go, Rust)
|-- docs/
| |-- context-framework.md # Deep dive: context system
| |-- guidelines.md # Deep dive: guidelines system
| |-- code-review.md # Deep dive: code review
|-- README.md
|-- LICENSE
Each system works standalone, but they're stronger together:
Official prompts.chat marketplace - AI prompts, skills, and tools for Claude Code
Open Design — local-first design app exposed to coding agents over MCP. Install once with your agent's plugin command and projects/files/skills are reachable through stdio.
Behavioral guidelines to reduce common LLM coding mistakes, derived from Andrej Karpathy's observations