Alfredvc

Author, review, verify, and package general @aharness/core finite state machines. Use when Codex needs to design or write an aharness FSM for any workflow, convert an informal process into typed states and transitions, choose between strict/open/choice/passive/embed/final states, create or review installable aharness FSM packages, or diagnose aharness FSM verifier and authoring errors.

Fixture sidecar research skill for sidecar integration tests.

Fixture skill for installed package tests.

Fixture skill for same-package child tests.

Fixture skill for dependency package tests.

Speak like a Caribbean pirate captain. Hearty, theatrical, with sea-faring metaphors and the occasional "Arrr!".

Discovery workflow for the rema CLI. Browse current offers and bonus missions, check weekly sales campaigns, find stores, and view transaction history. Use when exploring what's available at REMA 1000.

Runtime contract for the rema CLI. Covers authentication, output format, error handling, and invocation patterns. Use as foundation before any other rema skill.

Shopping list management workflow for the rema CLI. Create lists, search products by name, add items with GTINs, check off and manage items. Use when building or maintaining grocery shopping lists.

Explore the Open Food Facts database structure. Discover valid category, label, ingredient, brand, and allergen tags using facets, then use them as filters in product search. Use when you don't know the exact tag names to filter by.

Look up food products by barcode or search with filters. Find products by category, nutrition grade, eco-score, label, ingredient, or allergen. Use when you have a barcode or need to find products matching specific criteria.

Runtime contract for the openfoodfacts CLI. Covers output format, error handling, field filtering, and invocation patterns. No authentication required. Use as foundation before any other openfoodfacts skill.

Investigate a user's Claude Code setup to find ROOT causes of spend and recommend cost optimizations. Use whenever the user asks to "reduce my Claude cost", "why is Claude so expensive", "optimize my usage", "analyze my spend", "audit my setup", "what's driving my bill", "cost insights", or any variant of diagnosing their Claude Code habits and configuration. This is a diagnostic methodology skill — it sits on top of `cct-db` (which covers raw SQL) and guides the investigation itself so the agent doesn't declare victory on shallow findings. Trigger even if the user only mentions part of this (e.g. "my Opus bill is high" or "are my skills expensive?") — partial cost questions still need the full root-cause discipline.