Skip to content

DedInc/chef-agentic-skills / skills/chef-router

chef-router

Verified

Entry point for the Chef Agentic Skills library. Use first for general cooking, recipe creation or repair, meal planning, learning, adaptation, preservation, food-safety questions, and any request where several chef skills could apply. Route to the smallest useful chain and keep attention, experience, equipment and measurement needs separate.

FileKindSize
SKILL.mdskill md5.0 KB
agents/openai.yamlconfig93 B
---
name: chef-router
description: >-
  Entry point for the Chef Agentic Skills library. Use first for general cooking,
  recipe creation or repair, meal planning, learning, adaptation, preservation,
  food-safety questions, and any request where several chef skills could apply.
  Route to the smallest useful chain and keep attention, experience, equipment
  and measurement needs separate.
---

# Chef Router

The library is a brigade, not a checklist. Classify -> select -> read -> cook. Every selected skill must change the result.

## Shared contracts

- `chef-constraints-adaptation` defines cue-first, equipment-aware recipe writing. Read it when execution detail or equipment adaptation matters.
- `chef-food-safety` owns safety controls. **Read it when a safety number/process governs the answer**; do not treat an "always-on" label as if an unloaded table were already in context.
- `chef-accessible-cooking` is an explicit execution modifier for ADHD/СДВГ, fatigue, overwhelm, one-thing-at-a-time workflows, household measures, burner choreography or live stage-by-stage help. It counts as one routed skill and does not imply beginner technique.

## Step 1 — Primary verb

| Verb | Purpose | Usual lead |
|---|---|---|
| MAKE | create/cook a dish | named dish/cuisine skill, or `chef-recipe-developer` |
| FIX | rescue/audit | `chef-recipe-optimizer`, `chef-sauces-guide` or `chef-flavor-balance` |
| JUDGE | evaluate | `chef-culinary-inspector` |
| PLAN | menu/event/meal prep | `chef-menu-event-planning` or `diet-goal-nutrition` |
| LEARN | build skill | `chef-training-progression` |
| ADAPT | fit equipment/diet/substitution/execution | relevant owner |
| PRESERVE | ferment/pickle/can/cure | `chef-fermentation-preservation` |

## Step 2 — Material modifiers

| Signal | Add only when it changes the answer |
|---|---|
| iconic named dish with an actual bundled spec | `chef-world-classics` |
| Chinese/Japanese/French/Italian/Mexican/German/Indian/Silk Road context | one cuisine skill |
| meat / seafood / vegetables / grains-legumes as the main technical question | one ingredient guide |
| sauce failure | `chef-sauces-guide` |
| bread/dough | `chef-baking-doughs-guide` |
| pastry/chocolate/sugar | `chef-pastry-confectionery` |
| spices/blends/dosing | `chef-spice-guide` |
| appliance conversion | `chef-equipment-gastrophysics` |
| pantry improvisation | `chef-pantry-improviser` |
| unavailable ingredient | `chef-ingredient-localizer` |
| budget | `chef-kitchen-economics` |
| leftovers/scraps | `chef-zero-waste` |
| children/picky or food-sensory eating | `chef-kids-sensory` |
| allergy/restriction | one matching `diet-*` owner; add safety when cross-contact matters |
| ADHD/СДВГ, low focus, fatigue, one thing at a time, can't estimate by eye, household measures, burner-by-burner or explicit beginner execution help | `chef-accessible-cooking`; preserve the cook's actual technique level |

Negative triggers: an ADHD question unrelated to cooking is outside this library; a historical mention of ADHD does not force verbose execution for "what is cumin?"; "short answer" stays short. Adult attention support does not route to the kids skill unless food-sensory/child context is actually present.

## Step 3 — Chain discipline

Target **1–4 skills**, hard ceiling **5 per active route**. Complex projects may use separate bounded waves; do not create a 12-skill simultaneous context.

`chef-world-classics` wins only when it contains the named dish. A famous name is not proof that a spec exists. A cuisine skill usually beats a generic ingredient guide for cuisine-specific method; add the guide only for a missing hard number/failure mode. For live rescue, speed wins: one repair skill can be enough.

## Examples

| Request | Chain |
|---|---|
| "Курица по-китайски, СДВГ, без параллельной суеты" | `chef-cuisine-chinese` + `chef-accessible-cooking`; add safety if a chicken endpoint is needed |
| "Придумай мясо с рисом, новичок, нужны ложки и конфорки" | `chef-recipe-developer` + `chef-meat-guide` + `chef-grains-legumes-guide` + `chef-accessible-cooking` |
| "Я опытный повар с СДВГ, распиши порядок карбонары" | `chef-world-classics` + `chef-accessible-cooking`; no knife kindergarten |
| "Сметана сворачивается прямо сейчас" | `chef-sauces-guide`; immediate rescue first |
| "Какие специи к нуту?" | `chef-spice-guide` |
| "Можно ли этот рис есть завтра вечером?" | `chef-food-safety` |

## Output shape

MAKE -> executable recipe. FIX -> immediate cause/fix, full audit only if requested. JUDGE -> verdict + prioritized fixes. PLAN -> plan/timeline. LEARN -> next drill. ADAPT -> adapted artifact + tradeoffs. PRESERVE -> complete controlled method + source/safety gates.

Do not stack audit prose, legal boilerplate and a recipe unless the user asks. Project-level responsibility is in `/DISCLAIMER.md`; local safety warnings still appear exactly where they change an action.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall