Curate a large multi-source home-cooking recipe pool, classify it into planning-ready buckets, generate a balanced weekly meal plan, and render daily reminders. Use when the user wants recipe-pool building, home-cooking filtering, weekly meal planning, or daily menu delivery from GitHub cookbook sources.
Use this skill when the task is not "pick a few dishes right now", but "maintain a reusable recipe pool and plan a real week of home cooking from it".
The system is built around four layers:
For normal day-to-day use, treat this skill as a weekly plan + daily reminder system.
Default behavior:
If the user simply asks for today's meals or this week's meals, prefer the persisted week state instead of re-curating the whole recipe pool.
When another agent uses this skill, follow these defaults.
Do:
plan_week.pygenerate_weekly_plan.pyDo not:
Do:
plan_week.py only if the saved week is missing or stalegenerate_daily_plan.pyDo not:
Do:
/home/dsm/workspace/MealPlanner/diet/preferences.jsonPrefer:
exactTitles when the user names one specific dishtitleKeywords when the user means a dish family such as 鸡爪sourcePaths only when the exact source entry mattersDo:
rebuild_pipeline.pyUse the full rebuild path only for data/pool changes, not routine daily requests.
Default user profile:
Planning principles:
Supported cookbook repositories:
Anduin2017/HowToCookGar-b-age/CookLikeHOCSource policy:
HowToCook is the primary home-cooking sourceCookLikeHOC is supplemental and should be down-ranked when entries depend on chain-kitchen bases, prebuilt sauces, or semi-finished componentsGenerated data lives in /home/dsm/workspace/MealPlanner/diet/imported_recipes.json, /home/dsm/workspace/MealPlanner/diet/refined_recipe_pool.json, /home/dsm/workspace/MealPlanner/diet/structured_recipe_pool.json, /home/dsm/workspace/MealPlanner/diet/meal_pool.json, and /home/dsm/workspace/MealPlanner/diet/state.json.
Human-readable output lives in /home/dsm/workspace/MealPlanner/diet/weekly_meal_plan.md.
User preference overrides live in /home/dsm/workspace/MealPlanner/diet/preferences.json.
/home/dsm/workspace/MealPlanner/meal-planner/scripts/import_sources.py
Imports only recipe markdown from supported repositories./home/dsm/workspace/MealPlanner/meal-planner/scripts/refine_recipe_pool.py
Scores, filters, and deduplicates imported recipes./home/dsm/workspace/MealPlanner/meal-planner/scripts/build_structured_pool.py
Builds planning-ready grouped pools: mains, sides, soups, breakfast templates./home/dsm/workspace/MealPlanner/meal-planner/scripts/plan_week.py
Generates the current week's plan and persists it./home/dsm/workspace/MealPlanner/meal-planner/scripts/generate_weekly_plan.py
Renders the saved weekly plan./home/dsm/workspace/MealPlanner/meal-planner/scripts/generate_daily_plan.py
Renders today's reminder from persisted state./home/dsm/workspace/MealPlanner/meal-planner/scripts/rebuild_pipeline.py
Runs the full import -> refine -> pool build -> weekly plan -> render pipeline./home/dsm/workspace/MealPlanner/meal-planner/scripts/add_blacklist.py
Adds user blacklist rules into .If the user says a dish should never be suggested again, add it to /home/dsm/workspace/MealPlanner/diet/preferences.json and rebuild the pipeline.
Supported blacklist scopes:
exactTitles: exact recipe titlestitleKeywords: partial title keywordssourcePaths: exact imported relative pathsConvenient command:
python3 /home/dsm/workspace/MealPlanner/meal-planner/scripts/add_blacklist.py --title "宫保鸡丁的做法"
This is the default operational path the agent should prefer:
plan_week.py and generate_weekly_plan.pygenerate_daily_plan.pyOnly use the full rebuild path when:
Run when source repositories or filtering heuristics changed:
python3 /home/dsm/workspace/MealPlanner/meal-planner/scripts/rebuild_pipeline.py
Run when starting a new week or when the saved week is missing/stale:
python3 /home/dsm/workspace/MealPlanner/meal-planner/scripts/plan_week.py
python3 /home/dsm/workspace/MealPlanner/meal-planner/scripts/generate_weekly_plan.py
Run on normal days:
python3 /home/dsm/workspace/MealPlanner/meal-planner/scripts/generate_daily_plan.py
references/data-model.md: generated file layout and state modelreferences/selection-rules.md: weekly planning constraintsreferences/scoring-model.md: recipe filtering and ranking heuristicsreferences/integration-sources.md: source-specific import policyreferences/runbook.md: operational usageWhen the pipeline changes, update both code and these references together.
diet/preferences.jsonToast — restaurant POS, orders, menus, employees, revenue centers, and reporting.