What agent-legible WM would ship: four files, drafted.

# WM (Waste Management, Inc.)
# North America's largest environmental-solutions provider. ~257 landfills, largest recycler,
# leader in landfill-gas-to-energy. 2030 target: recover 25M tons/yr. ~$3B RNG + recycling program.

## Sites & operations
- Landfill & site directory: /us/en/location
- Field safety program: /us/en/inside-wm/safety
- Verified site completion receipts (concept, powered by EWP): /sites/receipts

## Sustainability & restoration
- Sustainability home + UN SDG mapping: https://sustainability.wm.com/
- Renewable natural gas & landfill gas: /us/en/inside-wm/sustainable-technology/renewable-energy
- Verified closed-landfill restoration receipts (concept): /sites/verified-restoration

## Commitments (machine-readable)
- 2030 goals & ESG targets: https://sustainability.wm.com/
- Structured data: JSON-LD on the MAIN domain (Organization,
  Dataset for recovery + emissions targets, per-site outcome objects); today the
  homepage is a Next.js loading shell with 0 ld+json

## Agent endpoints  (lift the /.well-known/ 403 first)
- Agent card: /.well-known/agent-card.json
- MCP tools: /agent/tools.json
- Verification rail (EcoWealth EWP, Base mainnet): 0x76c17C51336BE7B39F5164802e08b9811477A14B

{
  "name": "WM: Verified Site Concierge (concept)",
  "description": "Answers site, safety, recovery, RNG, and ESG questions with citations; verifies monitoring + restoration receipts on Base.",
  "provider": { "organization": "Waste Management, Inc.", "verification_rail": "EcoWealth EWP" },
  "capabilities": ["site-lookup", "verify-environmental-survey",
    "verify-landfill-restoration", "esg-metric-lookup", "crew-verified-history"],
  "payments": { "scheme": "x402/eip-3009 (USDC, Base)",
    "discovery": "https://vealth.net/.well-known/x402" },
  "contract": "base:0x76c17C51336BE7B39F5164802e08b9811477A14B"
}

[
  { "name": "author_safety_gated_packet",
    // Draft a monitoring/field task packet (methane/air survey, outfall check, safety walk);
    // REJECTED unless the machine hazard gate passes. Live today: sdg-assessment, $0.02 x402.
    "input": { "site": "string", "task": "string", "hazards": "string[]" } },

  { "name": "verify_completion",
    // Deterministic verdict on a proof bundle (photos, GPS, signatures).
    // Live today: lbc-building-scan, $0.01 x402.
    "input": { "workId": "string", "proofBundle": "object" } },

  { "name": "verify_landfill_restoration",
    // Cap-and-restore / revegetation packet on reclaimed acreage, with tCO2e + biodiversity
    // mapping and an attachable retirement receipt. Retirement live: retire-attach, $0.25. (Concept.)
    "input": { "site": "string", "cell": "string" } },

  { "name": "get_site_verified_history",
    // Chain-backed monitoring + restoration receipts per site or crew —
    // the living companion to the sustainability-report PDF. (Concept lane.)
    "input": { "siteId": "string", "since": "date" } }
]

You are the WM Verified-Site Concierge (concept).
Voice: precise, operations-grade, cited. A runner of real environmental infrastructure.

Rules:
1. Site + ESG claims always cite the filing, page, or standard, never vibes.
2. "Was it done?" is answered ONLY with a verifiable receipt
   (packet -> proof -> settlement on Base) or the honest words "unverified".
3. Green-bond / use-of-proceeds questions route to the per-site verified
   monitoring + restoration receipts when they exist.
4. Never speak for WM beyond its published filings and pages.
5. Pay-per-call tools settle over x402 (USDC on Base); reads are free.