| name | orchestrate |
| description | Full multi-LLM pipeline orchestration. Runs the recipe pipeline that coordinates GPT (lead), Gemini, Grok, and Claude for recipe generation and validation. Auto-detects recipe mode from this repository. |
Orchestrate — Multi-LLM Pipeline
GPT leads generation. Other models consult. Claude validates safety.
Usage
/orchestrate "task description"
/orchestrate recipe "Generate a classic Southern cornbread recipe with variations"
Mode is recipe by default in this repository.
How It Works
Pipeline defined in /home/user/ken/orchestrator/modes/recipe.yaml:
- Generate (GPT) — Structured recipe: ingredients, instructions, variations
- Expand (Gemini) — Ingredient knowledge, substitutions, nutrition, cooking science
- Safety Check (Claude) — Cooking temperatures, allergens, food safety, instruction clarity
- Creative Variation (Grok, optional) — Unexpected flavor combos, regional twists
Single pass — no loops.
Backend Invocation
IMPORTANT: Execute these commands directly using the Bash tool. Do NOT check if files exist first — just run them.
bash /home/user/ken/orchestrator/bootstrap-env.sh 2>/dev/null; pip3 install -q -r /home/user/ken/orchestrator/requirements.txt 2>/dev/null && python3 /home/user/ken/orchestrator/orchestrate.py recipe "task description"
Only if the command fails with No such file or directory or ModuleNotFoundError, tell the user:
"The orchestrator backend isn't available. Make sure the ken repo is cloned to /home/user/ken/ and run pip3 install -r /home/user/ken/orchestrator/requirements.txt."
Context Boundaries
SEND to external models
- Recipe requirements, ingredient lists, dietary constraints
NEVER SEND to external models
- Family attribution details, site analytics
Constraints
- Clear, step-by-step instructions
- SEO optimized (schema.org Recipe markup)
- Accessible formatting
- Safe cooking temperatures verified
- Allergen warnings where applicable
