Water Investment Navigator

View saved version →

---
name: utility-capital-plans
description: Answer questions about water utility capital improvement plans (CIPs) and planned capex — "which utilities have the biggest capital plans in Texas", "what is Houston planning to spend on water treatment", "find planned reservoir projects in California", "how has this project's cost changed" — using the Water Investment Navigator MCP capital-plan tools.
---

# Utility capital plans

Use the `water-investment-navigator` MCP tools. A capital plan (CIP) is a utility's published
multi-year budget of planned projects. **CIP amounts are planned spending, not
awarded contracts or money spent.** Don't present them as wins or as actuals.

## Which tool

| Question | Tool |
|---|---|
| Rank/screen utilities by planned capex | `latest_capex(state=..., min_total_musd=...)`: one row per utility, its newest plan |
| All plans for a utility, or older vintages | `capital_plans(utility_id=..., latest_only=false)` |
| One plan's totals, sector split, funding | `get_capital_plan(file_id, spending_group_by="sector")` |
| Projects in one plan | `capital_projects(file_id=..., sort="total_desc")` |
| Projects across utilities | `capital_projects(q=..., state=..., sector=..., min_total_musd=...)`: newest plan per utility by default |
| One project, cost drift, linked decisions | `get_capital_project(cip_id)` |

## Reading the numbers

- Every `*_musd` field is **millions of USD** (`source_currency` shows what the
  plan was published in).
- A plan covers `cip_start_year`–`cip_end_year`. `spending.by_year` is the
  yearly profile; `prior_years_musd` is spend before the window and
  `future_years_musd` is spend after it. Say which of these your total includes.
- `expenditure_status` (e.g. Ongoing, Complete) and `new_in_cip` flag
  whether a project is new in this vintage.
- `get_capital_project` → `versions` is the same asset across plan vintages. Use
  it for "how has the cost/timing changed". `decisions` lists the governing-body
  decisions linked to the project, which is the bridge from planned to awarded
  (use the water-contract-research skill from there).
- Plans are published at different times, so comparing utilities means
  comparing different plan years. Mention `publication_year` when ranking.

Cite the plan with its `file_name`, `publication_year` and `documentlink`
(a public link to the source PDF).

View saved version →

---
name: water-contract-research
description: Answer questions about what US water/wastewater utilities decided, awarded or approved, and which companies won or bid — "who won X", "what has firm Y won", "what did utility Z approve", competitor win/loss tracking. Uses the Water Investment Navigator MCP tools and cites the source meeting documents.
---

# Water contract research

Use the `water-investment-navigator` MCP tools. The data is decisions taken by utility governing
bodies (boards, councils, commissions), extracted from their agendas, minutes and
meeting video, with companies attributed to the contract events in each decision.

## Workflow

1. **Resolve the entities first.**
   - Utility named → `find_utilities(q=...)` to get `utility_id`. Several rows can
     match (a city and its water department); pick the one whose name fits the
     question, or say which you chose.
   - Company named → `find_companies(q=...)` to get `company_id`. Joint ventures are
     separate companies (e.g. "Florida Water Partners (Garney / Wharton-Smith)");
     include them when the question is about a parent firm, and say you did.
2. **List candidates.** `search_decisions` with the ids and filters (`state`,
   `date_from`/`date_to` by publication date, `min_value_millions`,
   `decision_type`), or `company_decisions(company_id)` for a firm's history.
   Follow `next_cursor` until you have the whole window; don't answer "all of X"
   from the first page.
3. **Read before you claim.** For each decision you report,
   `get_decision(decision_id, include=["companies","bid_tabulations","documents"])`.
4. **Quote the source** for any number or winner you state:
   `get_decision_source_text(decision_id)` gives the verbatim agenda/minutes text;
   `get_document_link(resource_id)` gives a 5-minute PDF link for a document with
   `has_pdf=true`.

## Reading the fields correctly

- `decision_type` is one of: Contract Awards, Contract Amendment, Capital Projects,
  Finance, Operations & Maintenance, Presentation/Report, People, Procedural, Other.
  Only Contract Awards / Contract Amendment are contract actions. A
  Presentation/Report decision with a big number is usually a plan or study, not
  money awarded.
- A company's `role` says what it did: `contractor`, `consultant`, `engineer`,
  `vendor`, `supplier`, `cmar`, `design_build`, `subcontractor`, `recipient` —
  and **`bidder`, which means it bid, not that it won.** For who won, look for
  a winning role or `is_winner` on the bid tabulation rows.
- `event_type` separates awards (`award`, `bid_award`, `task_order`, `gmp_set`)
  from `amendment`, `change_order`, `scope_change`, and from money that is not a
  won contract: `opex_payment`, `intergov_payment`, `finance_payment`,
  `financing`, `grant`, `acquisition`. `award_unverified` is a lower-confidence award.
- `evidence_role`: `final_vote` means the body acted on it in this decision;
  `reference` means the decision only mentions an earlier action.
- Money: `value_millions` is **millions of USD** and is null when the source
  gave a range or no figure. `delta_value` on a company event is in whole units
  of `value_currency`.
- Joint awards: when several firms share an award with no stated split, each
  firm is shown with the full value. Never add those across firms, and don't
  total a firm's wins across a JV and its members without saying so.
- `get_company` totals (`total_committed_value`) exclude opex payments.

## Answer format

Lead with the answer. For each decision: utility, date, what was approved, value,
winner/bidders, and a one-line quote from the source text. Say when data is
thin (few decisions, no bid tab) rather than filling the gap. The data refreshes
daily around 09:15 UTC and can change as quality review runs.

View saved version →

---
name: water-pricing-benchmarks
description: Answer unit-cost questions for water and wastewater construction and supply — "what does 12-inch PVC water main cost per foot", "is this bid price high", "typical price for a 36-inch gravity sewer" — from awarded and bid unit prices, using the Water Investment Navigator MCP pricing tools.
---

# Water pricing benchmarks

Use the `water-investment-navigator` MCP tools `pricing_benchmarks` and `pricing_observations`.

## Workflow

1. `pricing_benchmarks(q=...)`: search by item words (size, material, item type),
   e.g. `q="12 in PVC"`. Each row is one canonical item **per unit** (`LF`, `EA`,
   `CY`, `TON`, ...) with `observations`, `n_utilities`, `p25_unit_price`,
   `median_unit_price`, `p75_unit_price` and `quality_tier`. Match the unit the
   user asked about; the same item can appear once per unit.
2. Check `quality_tier`:
   - `benchmark`: enough observations across utilities to quote a range.
   - `indicative`: quote it, but say it is indicative and give the count.
   - `insufficient`: prices are null **by design**. Don't make up a number.
     Say there isn't enough data, and optionally list the individual observations.
3. `pricing_observations(canonical_item_id, unit=...)` shows the individual
   prices behind a benchmark: utility, date, bidder, `price_kind`
   (`awarded` vs bid/quoted), `is_winner`, quantity, and the `decision_id` to cite.
   Use it to explain spread (small quantities cost more per unit; older
   observations predate recent inflation) and to find comparable local prices.

## Answering

- Quote the median with the p25–p75 range, the unit, the observation count and
  the date range (`first_observed`–`last_observed`).
- Prefer awarded (winning) prices when the user asks what work "costs"; losing
  bids are useful for spread, not for what utilities actually pay.
- Prices are nominal and not adjusted for inflation or region. Say so when the
  observation window spans several years.
- Cite at least one source decision (`get_decision_source_text`) when a user
  will act on the number.

View saved version →

---
name: win-intelligence
description: Use Water Investment Navigator MCP tools to answer questions about US municipal water and wastewater investment, projects, utilities, capital plans and procurement decisions.
---

# Water Investment Navigator intelligence

Use the connected Water Investment Navigator MCP server as the primary source for WIN-specific market intelligence.

## Workflow
- Identify whether the user wants information on utilities, projects, capital plans, funding, procurement/decisions, companies, or market trends.
- Use the MCP tools to retrieve the most relevant current WIN data rather than guessing or relying on generic web knowledge.
- When possible, give the utility, project, decision, date, value and source context returned by WIN.
- Distinguish planned investment from awarded/procured work and from forecasts.
- For comparisons, state the population and time period covered by the returned data.
- If WIN data is incomplete or ambiguous, say so and ask a focused follow-up rather than inventing details.
- Treat WIN as the authoritative source for WIN-specific figures and records; use general knowledge only for context.

## Style
Be concise, commercially useful and factual. Highlight concrete opportunities, investment signals and procurement developments without hype. When summarising meeting or decision intelligence, preserve the distinction between what a public body decided, discussed or proposed.