/WastewaterCA

metadata:
  short-description: "Retrieve California wastewater surveillance for COVID, flu, RSV, and other pathogens, including activity levels, concentrations, and weekly trends."
name: WastewaterCA
version: 1.3.2
evals:
  manual:
    reason: "Surveillance answers depend on the requested pathogen, geography, and the current government datasets rather than a frozen fixture."
    falsifier: "The output reports a value for the wrong pathogen, site, region, or reporting week, or lacks the dataset reference needed to verify it."
description: "Pull California population-level pathogen surveillance from CDC NWSS and CDPH/CHHS Open Data — SARS-CoV-2, Influenza A/B, RSV, Mpox, Norovirus, H5, Candida auris from CA wastewater plants, statewide or for a configured region. Intents: GetCovidData (CDC viral activity level per site, week-over-week trend), GetWastewaterData (multi-pathogen panel). Leads clinical 4-7 days. USE WHEN wastewater, California COVID, viral load, COVID status, COVID levels near me, flu data, RSV, pathogen tracking, outbreak monitoring. NOT FOR national or non-CA surveillance."

California Wastewater Data Tracking

Pulls California population-level pathogen surveillance from two public sources: CDC's National Wastewater Surveillance System (a viral activity level per treatment plant for SARS-CoV-2, Influenza A and RSV) and CDPH's sample-level data on the CHHS Open Data Portal (concentrations for every pathogen target CDPH tests). Answers cover California statewide and, when configured, one region of counties.

Clinical case counts miss asymptomatic, unreported and untested infections and trail community spread by days. Wastewater catches everything that enters the sewer regardless of testing behavior, so it leads clinical signals by 4–7 days. Both sources publish on their own schedules; this skill pulls the latest figures each run and turns them into a level and trend you can act on before booking travel or an event.

Ideal State

• Every answer names the pathogen, the geography it covers (the configured region by name, or California statewide) and the latest data date, so no figure reads as real-time. The date comes from the dataset's own week or sample date.
• A COVID answer gives CDC's own activity level per site and the week-over-week direction; a panel answer covers every pathogen target the CDPH data carries in the past year.
• Figures come from a pull made in this run through `Tools/Wastewater.ts`, with the dataset named so the value can be checked.

Intent Routing

| Intent | Trigger | File | |--------|---------|------| | **GetCovidData** | "COVID numbers", "COVID status", "COVID data", viral load, activity level, trend before events/travel; default when ambiguous | `Intents/GetCovidData.md` | | **GetWastewaterData** | "wastewater", "respiratory viruses", "flu", "RSV", "norovirus", multi-pathogen panel | `Intents/GetWastewaterData.md` |

| File | Read when | |------|-----------| | `Tools/Wastewater.help.md` | flags, exit codes, what each command reads | | `Setup.md` | first-time setup, the region config, the proof command |

Data & Config

First-time setup, accounts and a proof command: [Setup.md](Setup.md).

• **Region:** `~/.claude/LIFEOS/USER/CUSTOMIZATIONS/SKILLS/WastewaterCA/Config.md` (or the file `WASTEWATER_CONFIG` names) holds `- region:`, `- counties:` as `Name FIPS` pairs, and an optional `- rpho region:` for the clinical layer. `bun skills/WastewaterCA/Tools/Wastewater.ts config` prints what a run resolves to.
• **Absent:** with no Config.md every command runs statewide and says so. A county line without a valid `06xxx` FIPS code stops the run with exit 2 rather than silently going statewide.
• No account, key or local dataset; the tool reads public APIs and writes nothing.

Gotchas

• **CDPH CSV dates are MM/DD/YYYY — never string-sort them.** A lexical compare once reported "Dec 31, 2025" as latest while the data ran into 2026. The tool queries the portal's datastore, where dates are timestamps; parse to real dates in any other path.
• **County columns are FIPS codes in CDPH, names in CDC.** CDPH `county_names` holds codes such as `06067`; CDC `counties_served` holds names. Config.md carries both so each source is matched on its own field.
• **WVAL is not a concentration.** It is CDC's unitless index with pathogen-specific category cut-offs; report CDC's category, never a threshold of your own.
• **The old CDPH statewide COVID file is archived** (last week 2025-08-02). Anything still reading `wastewater.csv` from the Respiratory Virus Weekly Report is reading year-old data.

Examples

**Example 1: COVID before travel**

User: "What are the COVID numbers before my trip?"
→ GetCovidData intent
→ Wastewater.ts covid, then covid --statewide: the week ending date, sites reporting, how many at each CDC level,
  the median site WVAL and its week-over-week direction, region and state side by side

**Example 2: Multi-pathogen panel**

User: "How are flu and RSV looking in wastewater?"
→ GetWastewaterData intent
→ Wastewater.ts panel: one row per pathogen target with its data date, geography, two-week level against the
  12-month range, trend and sample count

**Example 3: Local view with the clinical layer**

User: "Is flu actually up around here?"
→ GetWastewaterData intent, with Wastewater.ts covid --pathogen flu and the CHHS dashboard rows for the configured rpho region
→ The regional panel, per-site CDC flu levels, and regional test positivity