LazyCodex Documentation

The default React development stack consists of three tools:

All tools are strictly for development environments.

The Rust Programmer persona follows a 'production Rust in 2026' philosophy focused on explicit allocation, compile-time proof, and zero hidden costs. The development style is characterized by:

• Explicit allocation: Using arenas for hot paths and preferring &[T] or Cow in signatures.
• Compile-time proof: Maximizing const fn and using const { assert!(...) } for guards.
• Zero hidden cost: Avoiding hidden .clone() or .to_string() calls; using slice-based APIs.
• Type-encoded invariants: Using Newtype wrappers and type-state patterns to encode logic in the type system.
• Deterministic cleanup: Using scopeguard::guard or Drop for RAII and rollback logic.

The ultrawork mode enforces a strict workflow to ensure work is verified by observable evidence rather than agent reports. It follows these principles:

• Strict TDD: Follows the RED → GREEN → SURFACE → CLEAN cycle.
• QA Scenarios: Generates at least 3 realistic QA scenarios scaled to the task's risk.
• Real QA Channels: Uses actual interfaces for verification, such as HTTP calls, tmux, browser interaction, computer use, CLI stdout, or data diffs.
• Verification Gate: Implements a binding gate that loops until the work is genuinely verified by evidence.

The Python Programmer skill follows a 'Modern Python' philosophy: type-strict, stack-first, and async-correct. The core principle is to make illegal states unrepresentable by using the type system as a compiler.

Key Constraints:

• Parse, don't validate: Constructors must produce typed objects or raise errors; never pass unvalidated data deep into the stack.
• Frozen by default: Use @dataclass(frozen=True, slots=True) or Pydantic model_config = ConfigDict(frozen=True).
• No Any or object: Both are banned. Use Protocol for structural typing, TypeVar for generics, or explicit unions.
• No cast() or type: ignore: Redesign types to be correct instead of bypassing the checker.
• No broad except: Catch specific exceptions. Use # noqa: BROAD_EXCEPT_OK only at top-level boundaries (CLI/HTTP handlers) with logging and re-raising.

The codex-rules plugin injects project rules into the model context using Codex lifecycle hooks:

• SessionStart and UserPromptSubmit: Loads static project instructions once per session.
• PostToolUse: Watches for the apply_patch tool by default. When matched, it injects file-specific rules as additionalContext. Note that this is context-only and does not rewrite tool output.
• PostCompact: Clears the per-session injection cache after compaction to allow relevant rules to be reintroduced into the compacted conversation.

Supported Rule Sources (Auto-selected):

• CONTEXT.md
• .omo/rules/**/*.md
• .claude/rules/**/*.md
• .cursor/rules/**/*.md
• .github/instructions/**/*.md
• .github/copilot-instructions.md

The frontend skill is a router for all frontend, web UI/UX, and visual work, including building, styling, redesigning, React setup, performance audits, and visual QA.

Triggers: Use this skill when your request involves: frontend, UI, UX, design, redesign, styling, layout, animation, motion, interaction, micro-interaction, make it feel alive, premium, luxury, minimal, brutalist, Awwwards, DESIGN.md, mockup, React, Lighthouse, accessibility, WCAG, Core Web Vitals, looks generic, make it pretty, like X brand, lazyweb, or design research.

The Shopify-inspired design system is a 'dark-first' cinematic theme characterized by deep forest-teal undertones (not pure black), ultra-light display typography, and a singular high-energy Neon Green accent.

Key Visual Characteristics:

• Atmosphere: Dark-first with deep forest-teal undertones (#02090A, #061A1C, #102620).
• Accent: Neon Green (#36F4A4) used exclusively for focus rings and interactive highlights.
• Shapes: Full-pill buttons (9999px radius) are the primary interactive shape.
• Depth: Multi-layered box shadows and subtle inset white highlights create photographic depth.
• Imagery: Product screenshots should be embedded in dark UI contexts to match the surrounding darkness.

The ulw-loop is an expert goal orchestration agent designed for systematic work decomposition using ultrawork mode.

Role:

• Conducts orchestration while delegating specific tasks to right-sized subagents.
• Plans durable multi-goal work and fans out independent tasks in parallel.
• Performs manual QA on every result and records only proven evidence.
• Follows an outcome-first, evidence-bound, and atomic decision-making style.

Goal:

• Deliver every goal defined in .omo/ulw-loop/goals.json end-to-end.
• Prove every success criterion using captured observable evidence from real-usage scenarios (e.g., HTTP, Terminal, Browser, or Computer use).
• Note: Green test suites are supporting evidence but do not constitute proof of completion.
• Audit all progress (passes, fails, blocks, steering changes, checkpoints) in .omo/ulw-loop/ledger.jsonl.

The design system is characterized by a 'cinematic, tool-obsessed dark canvas'. It uses an absolute black void (#000000) as the primary background to create depth and focus. Key visual elements include:

• Void Canvas: Pure black (#000000) background, avoiding warm or gray tints.
• Accent Color: Framer Blue (#0099ff) used sparingly for links, borders, and ring shadows.
• Typography Style: High-impact, compressed headlines using GT Walsheim with extreme negative letter-spacing.
• Interactive Elements: Pill-shaped buttons (40px–100px radius) with no sharp corners.
• Glassmorphism: Frosted glass effects using rgba(255, 255, 255, 0.1) on dark surfaces.
• Product-Centric: High-fidelity product screenshots serve as the primary hero art.