pycsamt.agents.tooling#
pycsamt.agents.tooling#
ToolAgent — analysis and data/IO utilities from the web app’s tools
menu, exposed as Agent-Master tasks that return a compact table (rendered
as a monospaced block in chat) plus optional figures:
strike— geoelectric strike per station + rose/analysis figure,dimensionality— 1-D / 2-D / 3-D classification per station × period,validator— per-station EDI quality checklist,coords— transform station lat/lon to UTM easting/northing,elevation— enrich stations with elevation from an open web API,converter— re-export the survey to CSV / JSON / EDI on disk,batch_export— render a bundle of standard plots and save them,freq_editor— confidence-based frequency QC (drop / mask / recover),layered_model— build & preview a synthetic 1-D resistivity model.
freq_editor mutates the survey: it runs out-of-place and hands the edited
Sites back through AgentResult.data["corrected_sites"] so the chat’s
post-processing modal can apply it to the session or export it — the same
pathway used by the static-shift / denoise correction workflows.
layered_model is synthetic and needs no loaded data.
The coords tool is read-only. The elevation tool reaches an external
network service, and converter / batch_export write files to a
user-supplied folder — these are gated by the parameter modal (the user picks
the API / output folder and clicks Run before anything outward-facing
happens) and report exactly what they did.
Like PlotAgent it never calls an LLM; it
turns a validated parameter set into a result the Agent Master can display.
Classes
|
Strike / dimensionality / validator analysis tasks. |