Rust Cursor Agent Foundation

Executive Summary

This repository provides a Cursor-native AI agent development foundation for Rust UI applications built with Slint. It implements the architectural recommendations from the Rust AI Agent Development Foundation report:

• Cargo workspace with SOLID boundaries (core_domain, infrastructure, ui_application)
• Strict lint guardrails that force agents to write production-grade Rust
• 10 project-scoped Cursor skills using the Brain-Tool-Context triad (SKILL.md + scripts/ + references/)
• RPI workflow (Research → Plan → Implement) with persistent planning docs
• Multi-agent validation pipeline (Executor / Validator / Critic)
• Context survival hooks that preserve session state across compaction

Open this project in Cursor to get rules, skills, hooks, and prompts automatically loaded.

Installation

Prerequisites

• Rust toolchain (1.75+): rustup.rs
• Linux system libraries for Slint/winit (Wayland or X11):

  # Debian/Ubuntu
  sudo apt install libxkbcommon-dev libfontconfig1-dev

• Python 3 (for JSON scripts)
• Optional: cargo-audit for security skill (cargo install cargo-audit)
• Optional: agent-browser for headless UI validation (set AGENT_BROWSER_URL for wasm/web targets)
• Optional: Rust nightly for JSON test output (rustup install nightly)

Setup

git clone <repo-url> rust_tools
cd rust_tools
cargo check --workspace
./scripts/lint_and_fmt.sh

Usage

RPI Workflow

1. Research — Use @rpi-research or .cursor/prompts/01-research.md. Output goes to docs/planning_docs/implementation_research.md. No code edits.

2. Plan — Use .cursor/prompts/02-plan.md. Create docs/planning_docs/implementation_plan.md. Get approval before coding.

3. Implement — Use @rpi-implement with .cursor/prompts/03-implement-step.md. One atomic step at a time.

4. Review — Run ./scripts/orchestrate_review.sh. Critic review via @adversarial-review in a fresh session.

Quality Commands

./scripts/lint_and_fmt.sh       # Format + Clippy
./scripts/test_json.sh          # Tests with JSON output
./scripts/quality_gate.sh       # Full validation gate
./scripts/orchestrate_review.sh # Validator + Critic (required before merge)

Running the UI

cargo run -p ui_application

Invoking Skills

Reference skills in Cursor chat:

• @rust-core — new Rust modules and domain logic
• @rust-lint-hunter — fix compiler/Clippy errors
• @slint-ui-component — Slint layout and components
• @adversarial-review — Critic code review

See AGENTS.md for the full skill roster and multi-agent roles.

Project Structure

rust_tools/
├── core_domain/          # Pure business logic
├── infrastructure/       # Storage adapters
├── ui_application/       # Slint UI
├── .cursor/
│   ├── rules/            # Agent behavioral rules
│   ├── skills/           # 10 agent skills
│   ├── prompts/          # RPI reusable prompts
│   └── hooks/            # Context survival hooks
├── docs/planning_docs/   # RPI persistent memory
└── scripts/              # Validation and orchestration

Developer documentation for sub-reports and planning workflow: docs/planning_docs/README.md.