Steam Companion

Reusable Steam gaming companion for profiles, recommendations, reviews, and preference memory.

Use this skill for Steam-related requests about gaming profiles, owned/favorite/disliked/wishlist games, recommendation context, and review help.

Architecture

• manifest.ts: registration surface for OpenClaw Core.
• types.ts: domain models and contracts.
• memory.ts: repository abstraction plus in-memory and JSON implementations.
• service.ts: business rules only.
• tools.ts: tool definitions and thin handlers.
• prompts.ts: provider-agnostic prompts.
• constants.ts: names, version, and scoring defaults.
• index.ts: public export surface.

Rules

• Keep all business logic in service.ts.
• Keep transport-specific code out of the skill.
• Keep AI provider logic out of the skill.
• Use memory.ts only for persistence adapters and model helpers.
• Prefer structured recommendation and review context over free-form prose.
• Explain recommendations with clear reasons, tradeoffs, and confidence limits.
• When the Core fetches Steam Web API data, pass it in as a snapshot and let the service merge it.
• The skill never reads API keys directly; the Core resolves them via OpenClaw secrets.