<< All versions

Skill v1.0.0

currentAutomated scan100/100
open-science-pillars/hydrology/load-nwis
──Details
PublishedSeptember 27, 2026 at 03:47 PM
Content Hashsha256:75dad9ea16d08460...
Git SHA
──Files
Files (1 file, 3.0 KB)
SKILL.md3.0 KBactive
SKILL.md · 59 lines · 3.0 KB

version: "1.0.0" name: load-nwis description: "Load USGS streamflow from the Water Data API through dataretrieval's waterdata module with the volume gate: sites, parameters, windows parsed; approval status and qualifiers surfaced and segregated; provenance summary."


load-nwis

Bring NWIS records into the session safely. Works by slash command or conversationally ("load daily discharge for site 09380000, 2023").

Behavior, in order

  1. Parse and show back: sites, parameter codes, window, dv vs iv

cadence.

  1. Consult the bundle for this load first. Consult installed

knowledge concepts first, as the core consult-knowledge skill sets out, by the sites, parameter, and window in play; the nwis skill lists the concepts this plugin resolves to. Read the connector concept (endpoints, paging, the optional key, depth per collection, the rate limit) and the streamflow dataset concept (site-identifier typing, the approval status and qualifier vocabulary and their meaning, the rating curve and its revisions, daily vs continuous structure, the deprecated client) and the gotchas the request or downstream intent triggers (the provisional-data caution whenever the window reaches the trailing period; the regulated-gauge caution for any drought, low-flow, or trend framing). Restate what applies and cite each by path; do not carry these facts in this skill.

  1. Volume sanity before fetching (hard gate). Estimate rows (iv is

far denser than dv; rows scale with sites times window length) and apply the project local-config threshold. At or below: state the estimate and destination, proceed. Above: STOP, present the estimate and a smaller alternative (shorter window, dv instead of iv, fewer sites), and wait for explicit confirmation.

  1. Load with flags kept: fetch through dataretrieval's waterdata

module (get_daily for daily statistics, get_continuous for instantaneous values), with the approval status and qualifier columns retained; the accounting reported per the vocabulary the dataset concept defines; nothing dropped silently.

  1. Summary as provenance: sites with names, parameters and units,

window, row counts by approval status and qualifier, cadence, the request URL the client reports (never the key), whether a key was in the environment, dataretrieval's version, cache location, and the concepts consulted.

Must NOT

  • Never fetch above the gate threshold without explicit confirmation, on

any surface. (Hard gate: fires without consulting anything.)

  • Never fetch through dataretrieval's nwis module or the legacy

Water Services host; the connector concept records why and until when they answer.

  • Never restate a dataset rule this skill could consult: the qualifier

handling, the site-identifier typing, and the provisional and regulated-gauge cautions live in the NWIS concepts the nwis skill lists and are read from them per load, not carried here. That is what lets a revised gotcha change this skill's behavior without editing it.

All versions