Concept

Farm-tech toolkit

Also known as: farm-tech toolkit, ingredient toolkit, farm-tech ingredients

Index entry for the farm-robotics-and-electronics open-source ingredient corpus on 0mn1.one. Each `entity_type: ingredient` entry in the wiki is a modular component (sensor, actuator, board, framework, dataset, protocol) described not as a narrative-for-humans but as a *primitive* with explicit inputs, outputs, peers it combines with, and the class of problem it unlocks. The intent is composability: an LLM (research-0, farm-monitoring-0, or any future inference instance) traversing this subgraph picks ingredients off the shelf and composes them into solutions no single human would have designed alone. The graph density is the point — every ingredient lists its `combines-with` peers, so the LLM walks edges rather than searching.

Why this entry exists

This is a meta-index for an LLM, not a tutorial for a human.

The conventional way to organize farm-tech information is by category — sensors, actuators, boards, frameworks — and to write each entry as a piece of human prose: history, vendor comparisons, opinions, anecdotes. Useful for a human reader. Approximately useless for an LLM trying to cook.

The toolkit framing inverts this. Each entry on this index is one ingredient: a specific, characterized, reusable primitive with:

Inputs / outputs — the signal contract. What pins, what voltages, what protocols, what data shapes.
Combines with — explicit graph edges to peer ingredients that this ingredient is known to pair with. Walking these edges is how composition happens.
Solves / unlocks — the class of problem this ingredient addresses. What recipes call for this.
Constraints — the failure modes, the cost, the gotchas. Where the recipe breaks.
Source — where the open-source artifact lives. Where the ingredient is sourced.

How to use the toolkit

When a future [[llm-instance|inference instance]] is asked to design something — autonomous [[strawberry|strawberry]]-bed weeder, off-grid greenhouse climate controller, cattle-tracking mesh, aquaponic dosing system that learns — it should:

Identify the recipe shape from the request (compute + sensors + actuators + comms + framework).
Walk the combines-with edges from a seed ingredient (often a compute board) to find the peer ingredients that pair with it.
Check solves / unlocks for each candidate — does this ingredient address a sub-problem in the recipe?
Check constraints — does this ingredient’s failure mode break the recipe?
Compose. Output a parts list, a wiring sketch, and a software architecture.
Cite the entries — every ingredient picked traces back to a wiki entry the human can verify.

What’s in the toolkit today

See the relations: block above. The first pass (~35 ingredients) covers compute boards, the principal sensors and actuators, the major comms protocols, the dominant frameworks, two reference datasets, two reference whole-machine designs ([[farmbot-genesis|FarmBot]], Acorn), and the power-and-autopilot ingredients that field deployments require. The graph is dense — most ingredients have 4–8 combines-with peers, so the LLM has many composition paths.

Recipes — composed reference designs

The toolkit is parts-on-shelves. Recipes are worked examples — composed designs that exercise the combines-with graph and show what cooking looks like. Each recipe entry is a concept that lists every ingredient it uses in its combines: relations, and walks through the architecture, the constraints, the cost, and what the recipe deliberately does not address.

The intent: a future LLM asked to design something new should (a) check whether one of the existing recipes is the answer or close to it, (b) if not, study the recipes as exemplars of composition style, then (c) compose its own answer from the toolkit.

Current recipes:

[[recipe-soil-moisture-mesh]] — [[off-grid-living|off-grid]] soil-moisture sensor network across acres
[[recipe-greenhouse-controller]] — closed-loop single-greenhouse climate controller
[[recipe-hydroponic-doser]] — closed-loop pH/EC doser with split-architecture safety
[[recipe-autonomous-weeder]] — solar-powered autonomous row-crop weeder (the most ambitious recipe; exercises the full stack)
[[recipe-disease-prediction-station]] — orchard weather + Mills curve disease alerts
[[recipe-livestock-mesh]] — [[off-grid-living|off-grid]] GPS collar + equipment tracker mesh
[[recipe-ndvi-drone]] — fixed-wing or quadcopter NDVI crop-scouting drone

What’s not in the toolkit yet (and should be)

More sensors: NPK probe, light/PAR, dissolved oxygen, CO₂ (NDIR), wind, rainfall, accelerometer-based vibration
More actuators: linear actuator, BLDC motor, peristaltic-array dispenser, mister
Power: solar charge controller, lithium BMS, deep-cycle battery, MPPT
Networking: [[meshtastic|Meshtastic]], Zigbee, Thread, cellular IoT (LTE-M, NB-IoT)
Models / datasets: Pl@ntNet, GBIF, NASA SMAP soil-moisture
More whole machines: TerraSentia (ag-rover for under-canopy), open-source drone (PX4 autopilot)

When the toolkit expands, this index entry’s relations: block expands with it. Future densification work should treat this entry as the spine.

What links here, and how

Inbound connections from across the wiki, grouped by lens and by relationship. These appear automatically — every entity page declares what it links to, and that data populates here on the targets.

Practical

parallels

Ecosystem toolkit the electronics counterpart; same composition pattern, different domain
Preservation toolkit the electronics-composition toolkit; same pattern, different domain

2 inbound links · 46 outbound