Independent capability brief · Not affiliated · Passive public reads only · 2026-07-14
The Home Depot: the retail machine is elite. The agent and outcome layers are the open seats.
Four evidenced observations from Home Depot's public surface, read the way an AI agent reads it, and a concrete concept for the layer Path to Pro's own trajectory points at: a funded, proof-verified, on-chain-settled work packet that turns a trained tradesperson into a paid one an owner can trust.
Independence & method. Unsolicited concept work by EcoWealth Corporation, not requested, reviewed, or endorsed by The Home Depot. Every observation below is from normal public requests (curl with a standard browser user-agent, robots.txt/sitemap/well-known reads). No scanning, no probing, no authentication, no bot wall routed around: where the server returned 403, that is the finding, captured as served. Trademarks referenced descriptively. Every statistic is cited and dated; live counters override any number printed here.
Credit first: this baseline is elite
The Akamai edge + bot management in front of homedepot.com is genuinely best-in-class retail infrastructure: it is exactly why a $159.5B-revenue catalog (FY2024, 2,347 stores, the world's largest home-improvement retailer) stays up under adversarial load and card-testing bots. The Path to Pro program is real and substantial: free skilled-trades training (pathtopro.com), a Path to Pro Network matching graduates to hiring pros, and a Home Depot Foundation that has committed $10M+ to trades training. robots.txt is clean and specific (cart / checkout / search correctly disallowed). None of the findings below are "your site is broken": they are about a different reader (the AI agent) and a next layer (verified outcomes) that Home Depot's own investments already point toward.
The bottom line, in plain English
Let a buyer's assistant read the catalog → the shopper who wanted Home Depot can actually reach it
Today an AI shopping for someone gets "Access Denied" at the door, so a customer who specifically wanted Home Depot's catalog slips away to whatever the assistant can read; opening it to the good agents keeps that ready buyer.
An agent front door for Path to Pro and projects → the demand already knocking gets in
Career-seekers and project shoppers increasingly ask an AI first, and Home Depot's own program pages can't be read by one; a machine-readable menu of "price this project," "find a Pro," "start Path to Pro" routes that demand straight into Home Depot instead of stalling.
A portable proof-of-work record for graduates → a trained worker becomes a trusted, repeat-buying one
Path to Pro trains and matches, but a hiring Pro still takes completion on faith; a verified record that a job was done, done safely, and paid makes the graduate a better hire, and a better repeat Home Depot customer, turning the training spend into compounding value.
Four findings, each with its receipt
Finding 1 ★
To a non-browser agent, homedepot.com is a 403 wall, not a catalog.
A normal HTTPS GET with a standard browser user-agent gets HTTP 403 "Access Denied" (AkamaiGHost) on the homepage, on a product URL, and on a category URL, while robots.txt and the XML sitemap return 200. The bot management that protects the catalog also makes it invisible to any AI agent that isn't a full JavaScript-executing, cookie-bearing browser.
Why it matters: shopping is moving into agents (ChatGPT, Gemini, Claude, and retail-specific assistants). When a buyer's agent tries to price a project's bill of materials, homedepot.com answers Access Denied, so a shopper who specifically wanted Home Depot's catalog can't get to it through the assistant they're using. The catalog is superb; the door for the next class of buyer is shut.
Finding 2
The Path to Pro labor surface is reachable, and machine-illegible.
Unlike the walled storefront, the corporate and program surfaces load fine for an agent. But they carry no structured data at all: corporate.homedepot.com/ renders 0 JSON-LD blocks, and pathtopro.com is a client-rendered SPA whose /llms.txt path returns the full 1.1 MB app shell instead of a 404: an agent fetching it gets an empty JavaScript scaffold, not content or a course/job catalog.
Why it matters: Path to Pro exists to close a ~400,000-job trades gap. The program's own pages can't tell an AI career agent "here are the free courses, here is who's hiring, here is how to enroll" in a machine-readable way. The demand is being routed through a surface agents can't parse.
Finding 3
No agent front door exists on any Home Depot property.
Across all three surfaces we checked (the storefront, the corporate site, and the Path to Pro site) every standard agent-discovery path is a true 404. There is no llms.txt, no ai.txt, and no .well-known/agent-card.json telling an AI reader what Home Depot offers or how to transact with it.
Why it matters: the retailer that ships the first real agent front door in home improvement (a menu of "price this project," "find a Pro near me," "start Path to Pro" that any AI can read and act on) meets the next class of buyer where they already are. Today that seat is empty industry-wide, and Home Depot's catalog is the natural one to fill it.
Finding 4
Training connects to a job. Nothing connects the job to verified, paid outcome.
Path to Pro's flywheel is train → match → hire. The open link is the one after that: a trained worker completes real work, safely, and gets paid, provably. A hiring Pro today takes a graduate's completion on faith; there is no portable, machine-verifiable record that a specific job was done, done safely, and settled. That record is exactly what an AI hiring agent, and the graduate building a reputation, will want next.
Path to Pro public flow (from corporate.homedepot.com + pathtopro.com):free training → Path to Pro Network match → hired by a Pro → ? (no verifiable proof-of-work / proof-of-safe-completion object)Context: ABC, U.S. construction needs 349,000 net new workers in 2026, 456,000 in 2027; 40%+ retire by 2031.
Why it matters: the value of a trades on-ramp compounds when the work it produces is verifiable. A graduate with a chain of proof-verified, safely-completed jobs is a fundamentally better hire, and a better repeat Home Depot customer, than one with a certificate alone.
The concept: Build-ladder work packets on top of Path to Pro
EcoWealth runs the Ecological Work Protocol (EWP): a funded, proof-verified, on-chain-settled work-packet layer, live on Base mainnet today (contract 0x76c17C…A14B, source-verified; first packet settled end-to-end as workId 14). Work is framed as three ladders: Build (Living Building Challenge), Grow & Cook (Michelin), Restore (UN SDGs), step-1-anyone-today up to mastery. Pointed at Home Depot's world, the Build ladder is the natural fit:
1 · Shop-at-THD-shaped work. A Build-ladder packet (e.g. install high-efficiency insulation or a rainwater system on an LBC-track home) is a bill of materials Home Depot already sells and labor its Pro network already employs. Agent-readable product data feeds an agent-built BOM directly into the packet.
2 · Proof of safe completion. The graduate closes the packet with photo + GPS + signature; a deterministic proof-checker (agent-payable, $0.01/call) returns a verdict before payout. No proof, no settlement: pay follows the proof, and hazard-gated authoring means the task can't direct anyone into unsafe work.
3 · A receipt that compounds. Settlement writes a public receipt on Base. Each receipt appends to the tradesperson's verified work-and-safety history: a portable reputation an AI hiring agent can read, and the missing last mile of the Path to Pro flywheel.
Integration path: zero Home Depot system access required
The pilot deploys entirely on EcoWealth infrastructure first: the Agent Kit (see the kit page) runs read-only against public Path to Pro / product content, and one Build-ladder packet is funded and settled end-to-end on our live board, proving the concept before anything touches Home Depot's stack. No storefront change, no bot-wall change, no customer-data access.
Evidence log
#
Claim
Source
Observed
Re-verify
1
www homepage 403 Access Denied (AkamaiGHost) to browser UA
www.homedepot.com/
403 · 2026-07-14
curl -A "Mozilla/5.0…Chrome" -sI …/
2
Product + category URLs also 403
/p/310376127, /b/Appliances/…
403 each
curl -o /dev/null -w '%{http_code}' …/p/310376127
3
robots.txt + sitemap served 200 (crawlers invited)
/robots.txt, /sitemap/main.xml
200 each
curl -sI …/robots.txt
4
corporate home reachable, 0 JSON-LD
corporate.homedepot.com/
200 · 0 ld+json
curl -s … | grep -c 'application/ld+json'
5
pathtopro.com is JS SPA; /llms.txt = 1.1MB app shell (soft-404)
Step one is a receipt, not a meeting. For about $500 (a rounding error against Home Depot's $159.5B in FY2024 sales, and pocket change beside the Home Depot Foundation's $10M+ trades pledge) EcoWealth funds and settles one Build-ladder work packet in metro Atlanta (Home Depot's home community): a Path-to-Pro-shaped efficiency install, hazard-gated, closed with photo+GPS+signature proof. What comes back is the pitch itself: a permanent Base receipt (verified done · verified safe · verified paid) that a Path to Pro graduate carries as portable reputation, directly advancing the program's own mission to close the trades gap.
Want it cheaper and faster? Have any Home Depot AI agent make one $0.10 work-packet x402 call (x402: the standard that lets an AI agent pay per API call automatically) against our live endpoint right now: menu at vealth.net/.well-known/x402, contract 0x76c17C…A14B. The on-chain receipt lands in seconds; that receipt is the whole argument.
Inbound-only: nothing here is outreach; this first step is offered only if a Home Depot conversation opens. Contact: [email protected].
