Launching today

Meal Magic
Local-first cooking & meal skill for AI agents
1 follower
Local-first cooking & meal skill for AI agents
1 follower
A local-first cooking & meal-memory skill for Codex, Antigravity & Claude Code. Stored in local SQLite with pure Python (zero API keys). Tracks recipes, plans, allergies & real cooked/eaten history separately.





Free
Launch Team

AWS Activate CreditsCredits that fuel your startup’s next milestone
Promoted
Hey Product Hunt! 👋
I built Meal Magic because I wanted an AI cooking assistant that actually remembers my household's allergies, preferences, and what we really cooked or ate — without locking our data into a cloud SaaS, requiring extra API keys, or mixing up "we planned pasta" with "we actually ate pasta."
How it works:
The skill (SKILL.md + references) teaches Codex, Antigravity, or Claude Code how to format complete recipes, respect per-member dietary restrictions, and suggest varied meals.
The storage helper (meal_store.py) runs locally using only the Python 3.10+ standard library and SQLite. No network calls, no telemetry, no external dependencies.
Cross-agent shared profile: All three hosts can share the same local household database on Windows, macOS, and Linux, with transactional migrations (currently schema v8) and idempotent retry keys.
What we just shipped in v0.7.0:
Full recipe updates and safe deletion (recipe-remove keeps historical meal snapshots intact)
Date-range filtering on meal-list and read-only variety analytics via meal-stats
Selective recipe-import with dry_run conflict previews
A verified Windows shared-profile fast path + 175 automated tests across Ubuntu, macOS, and Windows
I'd love your honest feedback on the UX, the CLI/skill architecture, or anything you'd want to see next. Happy to answer any questions!