Europe-Camions

Build a local truck shopping tracker called RigWatch. Single user, runs on my laptop, no accounts, no cloud, no telemetry.

Stack, no substitutions: Node 20, TypeScript, Express, better-sqlite3, EJS templates, plain CSS. No React, no ORM, no Docker.

Data model in SQLite:
- vehicle: id, title, make, model, year, mileage_km, engine_power_hp, axle_config, euro_class, body_type, country, dealer_name, source_url, notes, status (watching | contacted | rejected | bought), created_at
- price_point: id, vehicle_id, amount_cents, currency, seen_on (date)
- alert_event: id, vehicle_id, kind (price_drop | price_rise | stale), message, created_at, seen (bool)

Features, all server rendered:
1. Add and edit a vehicle by hand, including pasting a source URL. Optional: if a URL is given, fetch the page server side and try to prefill title and price from Open Graph tags and JSON-LD only. If that fails, leave fields blank and say so. Respect robots.txt, honour a 5 second timeout, do not crawl beyond the single URL given.
2. Record a new price point for a vehicle at any time. Show a sparkline or a simple bar list of price history plus total change since first seen.
3. List view with filters: make, country, euro class, year range, mileage range, status. Sort by price, mileage, price per 1000 km.
4. Compare view: pick two to four vehicles, render a side by side table of every field plus a computed value score (price divided by remaining useful mileage, with the assumed end of life km configurable in .env).
5. A single command, npm run check, that re-fetches every vehicle with a source_url, records a new price_point when the parsed price differs, writes alert_events for changes and for anything untouched in 30 days, and prints a summary. No background scheduler, no email.
6. CSV import and export of vehicles so a spreadsheet stays the source of truth if I want.

Out of scope, do not build: messaging sellers, payments, user auth, public listing pages, scraping search result pages, translation, image hosting beyond storing one image URL string.

Config in .env: PORT, DB_PATH, DEFAULT_CURRENCY, END_OF_LIFE_KM. Commit a .env.example, never a .env.

Deliver: npm install then npm run dev serving on localhost, a seed script with 8 fake trucks, and a README with the three commands and a one paragraph note that this tracks listings found elsewhere and is not a marketplace.

# AGENTS.md — Build guide for Europe-Camions

## Project scope
A local watchlist app where you record trucks you are considering, track asking price changes over time, compare cost per kilometre and mileage, and get flagged when something you saved moves.

Catalogue verdict: no. The software here is a search form over a database of listings, and yes, an agent can build that in an afternoon. What it cannot build is the several thousand trucks that dealers actually bothered to upload, which is the entire product. A private clone launches with zero inventory and zero buyers, so it answers no queries and sells no trucks. The only honest personal build is a tracker that sits on top of listings you already found: watchlists, price history, diesel of a spreadsheet with better manners. Useful if you are shopping for one truck, worthless as a replacement.
Use the implementation prompt below to define the deliverable. Complete each phase's acceptance checks before extending the scope.

## Working agreement
- Inspect the repository and its existing instructions before choosing paths, dependencies or commands. Keep one coherent stack and explain changes to the proposed architecture.
- Plan a vertical slice that accepts a real input and produces the useful output described below. Persist only the state the prompt calls for; respect memory-only and upstream-managed workflows. Use fixtures only when they are clearly labelled.
- After scaffolding, document the actual install, development, check and build commands in README and keep them synchronized with the package or project manifest. Do not report commands as successful unless they ran.
- Work in small steps. At handoff, list implemented flows, checks actually performed, remaining blockers, and any credentials or provider setup the owner must supply.
- Do not publish, spend money, contact customers, delete source data or run irreversible migrations without the project owner's authorization.

## Prerequisites
- Node 20 and a terminal
- Listings you enter yourself, or feeds you have permission to fetch
- No account or cloud service needed

## Stack and architecture
- Stack, no substitutions: Node 20, TypeScript, Express, better-sqlite3, EJS templates, plain CSS. No React, no ORM, no Docker.
- Model the inputs, state transitions, and outputs named in this heavy truck and trailer classifieds marketplace prompt; keep source IDs and timestamps.
- Interface for this heavy truck and trailer classifieds marketplace workflow: a local web page with input, progress, review, and export views.

## Security and data integrity
- Add and edit a vehicle by hand, including pasting a source URL. Optional: if a URL is given, fetch the page server side and try to prefill title and price from Open Graph tags and JSON-LD only. If that fails, leave fields blank and say so. Respect robots.txt, honour a 5 second timeout, do not crawl beyond the single URL given.
- Validate imported records and file paths; keep user data within the storage location named in the prompt.
- State the practical limit: The inventory: thousands of vehicles from dealers across several countries.

## Agent implementation rules
- Project rule, data: Model the inputs, state transitions, and outputs named in this heavy truck and trailer classifieds marketplace prompt; keep source IDs and timestamps.
- Project rule, behavior: vehicle: id, title, make, model, year, mileage_km, engine_power_hp, axle_config, euro_class, body_type, country, dealer_name, source_url, notes, status (watching | contacted | rejected | bought), created_at.
- Project rule, recovery: Add and edit a vehicle by hand, including pasting a source URL. Optional: if a URL is given, fetch the page server side and try to prefill title and price from Open Graph tags and JSON-LD only. If that fails, leave fields blank and say so. Respect robots.txt, honour a 5 second timeout, do not crawl beyond the single URL given.

## Optional agent skills and references
No installable skill is listed for this guide; use the project rules and the stack documentation.

Read the linked SKILL.md and its dependencies before adding a skill. Select only the skills matching this project's runtime and task; their documentation does not supply API access, credentials or approval to perform external actions. Pin the reviewed revision where the tool supports it. Follow the chosen agent's documented project-level installation mechanism.

## Distribution ideas
These are optional planning notes. Obtain the owner's approval before publishing or contacting anyone.
- Demonstrate this heavy truck and trailer classifieds marketplace operation with sample or user-owned data: vehicle: id, title, make, model, year, mileage_km, engine_power_hp, axle_config, euro_class, body_type, country, dealer_name, source_url, notes, status (watching | contacted | rejected | bought), created_at.
- Show a concrete result or export: CSV import and export of vehicles so a spreadsheet stays the source of truth if I want.
- State the practical limit: The inventory: thousands of vehicles from dealers across several countries.

## Engineering roadmap
1. Phase 1, architecture and data: Stack, no substitutions: Node 20, TypeScript, Express, better-sqlite3, EJS templates, plain CSS. No React, no ORM, no Docker. Model the inputs, state transitions, and outputs named in this heavy truck and trailer classifieds marketplace prompt; keep source IDs and timestamps.
2. Phase 2, implement: vehicle: id, title, make, model, year, mileage_km, engine_power_hp, axle_config, euro_class, body_type, country, dealer_name, source_url, notes, status (watching | contacted | rejected | bought), created_at.
3. Phase 3, implement: price_point: id, vehicle_id, amount_cents, currency, seen_on (date).
4. Phase 4, review and output: alert_event: id, vehicle_id, kind (price_drop | price_rise | stale), message, created_at, seen (bool). CSV import and export of vehicles so a spreadsheet stays the source of truth if I want.
5. Phase 5, recovery and acceptance: Add and edit a vehicle by hand, including pasting a source URL. Optional: if a URL is given, fetch the page server side and try to prefill title and price from Open Graph tags and JSON-LD only. If that fails, leave fields blank and say so. Respect robots.txt, honour a 5 second timeout, do not crawl beyond the single URL given. Verify this invariant with a saved fixture: An invalid input or interrupted operation must retain the source and show a recoverable state; exported records must reload with the same IDs. State the practical limit: The inventory: thousands of vehicles from dealers across several countries.

## Paid-product capabilities outside this build
- The inventory: thousands of vehicles from dealers across several countries
- The buyer side, so nothing you list gets seen
- Dealer vetting and the loose trust layer that comes with a known marketplace
- Multilingual reach across French, German, Dutch and Spanish speaking buyers
- Cross referencing by make, axle configuration, euro emission class and body type on real data

## Implementation prompt
Build a local truck shopping tracker called RigWatch. Single user, runs on my laptop, no accounts, no cloud, no telemetry.

Stack, no substitutions: Node 20, TypeScript, Express, better-sqlite3, EJS templates, plain CSS. No React, no ORM, no Docker.

Data model in SQLite:
- vehicle: id, title, make, model, year, mileage_km, engine_power_hp, axle_config, euro_class, body_type, country, dealer_name, source_url, notes, status (watching | contacted | rejected | bought), created_at
- price_point: id, vehicle_id, amount_cents, currency, seen_on (date)
- alert_event: id, vehicle_id, kind (price_drop | price_rise | stale), message, created_at, seen (bool)

Features, all server rendered:
1. Add and edit a vehicle by hand, including pasting a source URL. Optional: if a URL is given, fetch the page server side and try to prefill title and price from Open Graph tags and JSON-LD only. If that fails, leave fields blank and say so. Respect robots.txt, honour a 5 second timeout, do not crawl beyond the single URL given.
2. Record a new price point for a vehicle at any time. Show a sparkline or a simple bar list of price history plus total change since first seen.
3. List view with filters: make, country, euro class, year range, mileage range, status. Sort by price, mileage, price per 1000 km.
4. Compare view: pick two to four vehicles, render a side by side table of every field plus a computed value score (price divided by remaining useful mileage, with the assumed end of life km configurable in .env).
5. A single command, npm run check, that re-fetches every vehicle with a source_url, records a new price_point when the parsed price differs, writes alert_events for changes and for anything untouched in 30 days, and prints a summary. No background scheduler, no email.
6. CSV import and export of vehicles so a spreadsheet stays the source of truth if I want.

Out of scope, do not build: messaging sellers, payments, user auth, public listing pages, scraping search result pages, translation, image hosting beyond storing one image URL string.

Config in .env: PORT, DB_PATH, DEFAULT_CURRENCY, END_OF_LIFE_KM. Commit a .env.example, never a .env.

Deliver: npm install then npm run dev serving on localhost, a seed script with 8 fake trucks, and a README with the three commands and a one paragraph note that this tracks listings found elsewhere and is not a marketplace.

## Completion evidence
Demonstrate the prompt's acceptance scenarios against the scoped workflow. Include setup from a clean checkout and failure recovery. Check persistence across restart and export/restore only for the state the prompt says to store; for memory-only tools, confirm that temporary content is discarded as specified. Record actual results and remaining limitations. A detailed plan alone does not establish a working replacement.