Local Crop Recommendations Prompt

[silent wyvern]

Make a prioritized list of 12 crops as expert local gardener & farmer. Choose crops to maximize (calories, fat, protein, carbs, vitamins, color, fiber, market value) & minimize (pest/disease susceptibility, time to harvest), override with inputs from block after this prompt if given. Choose only crops suitable & optimal to plant on {specific date} in {specific location - city & state or region}. Omit filler words & use active voice throughout. Begin w/ bullets detailing:

Local Conditions:

• Date, Season: {Date & specific [early/mid/late] planting season}

• Location: {location}

• USDA Zone & Climate: {USDA Zone & local climate with avg temps & rainfall; no wolfram links}

• Pests & Diseases: {Common pests & diseases}

• Soil: {Soil type/quality}

• Priority: {paraphrase priority, for example: higher calories, fat, protein carbs, vitamins, color, & fiber, w/ low time to harvest & pest/disease susceptibility}:

Recommended Crops:

[NOTE: sequence entry format - newline before each]

{sequence_index} - {crop_name} (genus_species):

• High in : {relevant priorities}
• Harvest: {harvest in {n} days, around {specific_date}}
• Guide: {crop preferences: Plant in {soil} with {sunlight} {spacing} apart. Water {frequency} for {duration/intensity}.
• Pests/Diseases: {What are probabilities this crop is susceptible to pests or diseases? If none known, skip this point; Else, specific names of known pest/disease susceptibilities associated w/ specific names of resistant varieties &/or 'control with' methods}
• Seeds: {2+ reliable seed vendors}

Avoid bold, can't copy it. Avoid line breaks in bullets or headers. Each crop must be suitable to plant on given date & location. Don't repeat crops. Use wolfram 1x if available for "climate of {location}" data only. Don't use wolfram for crop data.

--- Arguments
Date =
Location =
Maximize = (calories, fat, protein, carbs, vitamins, color, fiber, market value)
Minimize = (pest/disease susceptibility, time to harvest)
--- Response

[tame pier]

Hmm, just ran this. Not sure of the purpose. The local conditions part is ace. Not sure about the recommended crops, formatting wonderful, but I'm like heh for the recommendations (I have an ag/hort bs). Hard to parse usefulness w/o specifics around acreage(/sq feet) and other limiting factors. I ran something on Quincy, Illinois and the first recommendation was corn but no soybeans (?) which makes me wonder too about how it is actually measuring the metrics (since like by the metrics it is using it should include soya). the seed vendor part seemed accurate. Good starting point for a more honed in focus!

Like market value might be a strange value to have since that is so based on where you are selling? Green beans are great, unless everyone else is selling green beans or all the stores in the area already partner with a green bean farmer (lol). Also a complete lack of perennial recommendations (which like understandable) makes me wonder about long term...curious about what happens if actual coordinates are used and if/how difference would show up (ie put the field on a northside riparian zone vs a southside prairie)

[silent wyvern]

You can definitely fiddle with the arguments, the order matters a lot. It might have placed corn before soybeans because calories are mentioned before protein in the priority order, if you put protein first I bet it would highlight beans higher in the list. Feel free to remove the market value consideration if you don’t like it, and you can also add the acreage information to the section on the sequence entry format. Admittedly I made it more for testing and I’m not an ag major or farmer — you can potentially make it more focused on the needs of farmers versus gardeners. Perhaps you can help figure out a way to use a plug-in to get up to date local market prices and we can make a new version of the prompt with all of these upgrades! Thank you so much for the feedback @tame pier !

[tame pier]

oh yeah, I was just giving some discussion! find it strange in the 12 it didn't mention soy at all! definitely an interesting starting point! curious how directed resources/appendixes can build on the model as well as stability in longer convos

[silent wyvern]

You could try nesting it inside scan.md (in prompt library) as the carry prompt, it’ll rewrite everything every iteration so you won’t lose context. Might be too long tho