---
name: fleet-overview
description: Answer "how is my fleet doing?" from Pulpo. Vehicle counts, counts by status or type, the fleet health score (Salud de flota), vehicles missing basic data, vehicles without driver, odometer, or insurance, expired or expiring insurance, vehicle alerts, and a short list of what needs attention. Read-only. Use for fleet size, fleet health, findings, and "what should I look at today".
---

# Pulpo fleet overview and health

Reads only. Everything is scoped to the selected account and to the segments the user can see (see `getting-started` if an account must be selected).

## Know which number is which

| Question | Tool | Notes |
| --- | --- | --- |
| How many vehicles in the account | `pulpo_get_vehicles_count` | Account-wide, no filters. |
| How many match a condition | `pulpo_list_vehicles` with `filters` and `take: 1` | Read the total from `_metadata._total_rows`. By default it covers the user's segments and excludes archived vehicles. |
| The fleet's health score | `pulpo_get_fleet_score` | Whole number 0 to 100, the average over the vehicles the user can see. `fleetScore: null` means nothing is scored yet; report `vehiclesEvaluated` out of `vehiclesTotal`. |
| One vehicle's score and what it lacks | `pulpo_get_vehicle_score` | Takes a Pulpo ID (`V-0012`) or plate; VIN or economic number also work. |
| What basic data is missing | `pulpo_get_fleet_basic_data_pending` | Counts per missing field (plate, VIN, brand, model, year) and the empty "En Alta" records, with the worst vehicles first. Filter with `only`, page with `skip`/`take`. |
| Findings on one vehicle | `pulpo_get_vehicle_fleet_health` | Needs the numeric `vehicleId`. Returns `discoveries`, which are findings, not the score. |
| Which findings are switched on | `pulpo_get_fleet_health_enabled_findings` | Call it before interpreting findings. |
| Active alerts on a vehicle | `pulpo_get_vehicle_alerts` | Numeric `vehicleId`. Empty array means none. |

Rules that save you from wrong answers:

- **The score is the score.** Quote `pulpo_get_fleet_score` as the screen shows it. Never compute your own average from findings or from list rows.
- **Never count missing data by paging `pulpo_list_vehicles`.** Its filters do not see empty fields and the numbers will not match the screen. Use `pulpo_get_fleet_basic_data_pending`.
- **Counts that disagree are not an error.** The account count and a filtered list can differ because of segments and archived vehicles. State the scope you used ("vehicles you can see, not archived").
- **A disabled finding is not a clean result.** If a finding type is off in `pulpo_get_fleet_health_enabled_findings`, say it is not being checked.
- Fleet health findings are `VEHICLES_IN_INOPERABLE_STATUSES`, `VEHICLES_WITHOUT_ASSIGNED_DRIVERS`, `VEHICLES_WITHOUT_RECENT_ODOMETER`, `VEHICLES_WITH_NEAR_ASSIGNMENT_EXPIRATION`, `VEHICLES_WITH_NEAR_INSURANCE_EXPIRATION`, `VEHICLES_WITHOUT_INSURANCE`, `VEHICLES_WITH_INSURANCE_EXPIRED`. There is no single call that lists a finding across the whole fleet, so build the fleet-level view from filters below and confirm the few vehicles that matter with `pulpo_get_vehicle_fleet_health`.

## Filters for the common questions

`pulpo_list_vehicles` takes `filters` as serialized JSON: `{"AND":[{"parent":"vehicle","<field>":{"<operator>":<value>}}]}`. Operators: `is`, `in`, `contains`, `range`, `gt`, `gte`, `lt`, `lte`, `empty`, `not`. Useful fields: `status` (status name), `vehicleStatus` (status id), `type`, `vehicleType`, `brand`, `year`, `archived`, `segments`, `assignedDriver`, `hasDriverAssignment`, `insuranceEndDate`, `insuranceSupplier`. Include `{"all": null}` only when the user wants every vehicle regardless of segment; include `{"archived":{"is":true}}` to look at archived ones.

- Vehicles by status: one count per status name the user cares about, or list a page and group it. Operable vs inoperable comes from the status catalog (`pulpo_list_catalog_items` with `className: "vehicles-status"`, the `behavior` field).
- Without a driver: `hasDriverAssignment` false. Insurance ending soon or expired: `insuranceEndDate` with a `range` from today to the horizon, or up to yesterday for expired. If a filter errors or returns something odd, read one vehicle with `pulpo_get_vehicle_details` and adjust to the real field names rather than guessing again.
- Odometer without a recent reading: use `pulpo_get_odometers_latest`, and ask what "recent" means to the user if they did not say.

## A good overview

1. Call `pulpo_get_fleet_score`, `pulpo_get_vehicles_count` and `pulpo_get_fleet_basic_data_pending` (independent, run together). Add the filtered counts the question needs.
2. Lead with the numbers: vehicles, score with how many are evaluated, what is missing.
3. Then at most five attention items, ordered by consequence for the operation: inoperable vehicles, expired insurance or contracts, no driver, no recent odometer, missing basic data. Name vehicles by plate, name or Pulpo ID, and say how many more there are.
4. Say what you did not check (a finding that is off, segments you cannot see, a page cut short).
5. Offer the next step, not a data dump: open one vehicle, check its alerts, list upcoming maintenance (`maintenance-reminders`).

Fixing what the overview finds (filling data, changing status, assigning a driver) is done with the `vehicles` and `drivers-assignments` skills, with their write rules.

## If the user asks you to fix something

This skill only reads. When the user says "fix it", switch to the skill that owns the change and follow its write rules, which are always the same: ask first and show exactly what will change (record by plate or name, current value to new value), wait for a plain yes to that exact action, never invent ids or values, and remember that deletes, archives and money movements cannot be undone.