OPEN FIELD TOOLS / 001
Make the link.
Show the work.
A radio budget can pass while the path still fails.
Find the difference before you raise the mast.
and community networks
THE WORKSPACE
From guesswork to a defensible plan
Your radio link
Examples are hypothetical design inputs, not field measurements
Radio budget
A / SIGNALPreparing the local calculator…
Every dB accounted for
A passing budget is a model result, not proof of a working link
What is in the way?
B / CLEARANCEUse one common elevation datum for both antennas and the obstruction top
ClearPath verdict
PORTABLE PEERSignal checked. Path unchecked.
Load a second agent network to screen the Fresnel zone around your line of sight.
ClearPath is an included companion built for this project. No external organization or real-world inspection is implied.
Screening only · verify the site, equipment and applicable spectrum rules
Find the smallest change that works
Use the current radio inputs. The solver enumerates the candidates below; the objective is explicit.
Antenna and power candidates
No unnecessary power. No hidden objective.
Every candidate is checked against the same radio equations and your supplied constraints. Physical clearance is assessed separately.
Three controlled alternatives
One change at a time: +4 dBi at both antennas, +3 dBm power, or half the distance
Run the comparison to see tradeoffs and EIRP flags.
How much distance reserve?
Borrow the expert. Keep the execution.
Public graph → validated tools → your browser. A peer cannot select your model or receive a key.
Choose a clearance network
The bundled peer works immediately. To test another origin, supply a network URL whose final component is its exact network name.
No third-party contact without approval
A different origin prompts before every fetch. Only public files are fetched; the task runs locally. Public artifact requests omit credentials.
Inference is a visitor choice
Default: browser symbolic planner / structured-task-v1
The default routes explicit typed goals through the published graph and computes real results. It is a limited deterministic model, not a generative LLM.
Optional OpenAI-compatible inference
Live execution ledger
Ordered agent calls, destinations, consent and errors. No keys or private-context values are recorded.
Exact execution events (JSON)
[]
A decision you can hand over
Save scenarios on this device, or export a self-contained input and result record with a SHA-256 digest.
Saved on this browser
The digest and numeric inputs are checked. Imported claims are discarded and radio results are recomputed; run the path audit again.
What the record proves
- Which numeric inputs were used
- Which equations produced each result
- Which constraints passed in the model
- Which portable network and tool integrity were used
It does not prove measured coverage, uptime, terrain accuracy or permission to transmit.
A radio budget pass and a single-obstruction clearance pass are preliminary screens. Neither establishes throughput or availability. All example scenarios are hypothetical. Read the equations and limits ↗
BUILT TO BE BORROWED
Humans get a workspace.
Agents get the same science.
Load the public network, inspect its instructions and schemas, and run its tools in your own caller runtime. No central registry or private API integration required.
Agent manifest ↗ Portable network ↗ Capability schemas ↗{
"capability": "explain_term",
"input": {
"term": "fresnel"
}
}Open the independent caller console →