Tethered Trust

What Tethered Trust is, what it covers, where its data comes from, pricing and terms.

无需输入参数。

Which product categories and models are in the compatibility graph, with counts. Pass a category to list its models. Covers anything that has to fit, plug in, mount, run on or charge with something someone owns. Verified catalog (specs checked against cited manufacturer sources): computers & consoles (RAM, M.2 SSDs, USB-C chargers), phones/tablets/watches (USB-C PD, MagSafe, Qi2, watch bands), cameras (lens mounts, adapters, batteries, memory cards), cordless power tools (battery platforms), printers (ink, toner, drums), appliances (refrigerator water and air filters, dryer vents and power cords, installation fit), vacuums (bags, filters, batteries, robot-vacuum parts), air quality (purifier filters, HVAC media, humidifier pads), e-bikes (batteries and chargers by drive system), smart home (Matter, Thread, Zigbee, hubs, bulb bases), PC building (CPU sockets, cooler mounts, memory, clearances, power), cars & EV charging (tire sizes by trim, NACS/J1772/CCS1 and approved adapters, home chargers; a starter set of models) and TVs, consoles & streaming (VESA mounts, weight and screen size, console generations, app platforms). Products outside the catalog in any category are read by AI in terms of the same standards and returned as unverified.

Checks a cart at checkout. Takes the checkout session exactly as a commerce protocol returns it (a UCP checkout or an ACP CheckoutSession) or plain line items, plus what the customer already owns. Returns a decision per line item, safe_to_checkout, an action (proceed, resolve_conditions, do_not_checkout, ask_customer), questions, substitutions, a signed receipt, and a com.tetheredtrust.shopping.compatibility extension object to attach to the checkout.

For retailers and marketplaces (API key required): up to 50 products per call (SKU with title, URL or GTIN) in; each product's compatibility attributes out (standards it takes and provides, numeric limits, evidence level, freshness), ready to add to a product feed or UCP catalog so agents can match it to what shoppers own.

Call this when the person gives a shop link, ASIN, barcode or vague product name and you need the exact model and what it's compatible with. Identify a product from a URL, model number, SKU, ASIN, UPC/EAN or description, and get its compatibility profile: the standards it takes and provides (mounts, sockets, connectors, platforms, part numbers, sizes), numeric limits (OS version, watts, weight, screen size), evidence and freshness. Products not in our reviewed catalog are researched on demand from the manufacturer's page and returned with status read_from_manufacturer_page (unreviewed) or unconfirmed (not enough to decide on). Use first when the product could be anything.

Every compatibility standard family (with reviewed values), the open formats used for new values, and the numeric limit keys. How Tethered Trust represents physical connections, dimensional fit, vehicles, software and services.

无需输入参数。

Look up one product in the Tethered Trust graph by name or model number and get its structured specs, what it accepts (mounts, battery platforms, cartridges, protocols) and what it fits. If the text could mean several models, returns them all as candidates (ambiguous: true) with ambiguous: true.

Call this before you recommend or order two or more items that have to work together or with what the person owns (a PC build or upgrade, a camera kit, a smart-home setup, a charger plus cable). Check a whole purchase before recommending or ordering it: the setup the person owns (with current configuration) plus everything in the cart. Returns one decision (supported / conditional / incompatible / insufficient_evidence), a result per part, configuration limits across the plan (memory slots and capacity, M.2 slots), adapters or hubs still needed, substitutions for parts that won't work, the exact questions blocking a decision, evidence, and a signed receipt. Covers anything that has to fit, plug in, mount, run on or charge with something someone owns. Verified catalog (specs checked against cited manufacturer sources): computers & consoles (RAM, M.2 SSDs, USB-C chargers), phones/tablets/watches (USB-C PD, MagSafe, Qi2, watch bands), cameras (lens mounts, adapters, batteries, memory cards), cordless power tools (battery platforms), printers (ink, toner, drums), appliances (refrigerator water and air filters, dryer vents and power cords, installation fit), vacuums (bags, filters, batteries, robot-vacuum parts), air quality (purifier filters, HVAC media, humidifier pads), e-bikes (batteries and chargers by drive system), smart home (Matter, Thread, Zigbee, hubs, bulb bases), PC building (CPU sockets, cooler mounts, memory, clearances, power), cars & EV charging (tire sizes by trim, NACS/J1772/CCS1 and approved adapters, home chargers; a starter set of models) and TVs, consoles & streaming (VESA mounts, weight and screen size, console generations, app platforms). Products outside the catalog in any category are read by AI in terms of the same standards and returned as unverified.

When a plan comes back incompatible or conditional: the smallest set of changes that makes it work (swap a part for a verified one that fits everything else, add a missing cooler, hub or adapter, trim quantities that exceed slots), re-validated, with the reason for each change. Fewest changes and verified products first; no prices.

What can be upgraded on this device and with what: memory (type, slots, maximum), storage (M.2 slots, lengths, PCIe generation), power, CPU and cooling for motherboards, graphics card and cooler clearance for PC cases, plus the parts in our graph that fit (verified first) and the manufacturer's notes.

Read the brand, model and part numbers from a photo of a label or nameplate and match them to products. Matches are returned as candidates, with what was read. The photo isn't stored.

Re-run a saved purchase plan against today's data before ordering, and see what changed since it was issued: decisions, product records, or receipt status.

Up to 25 independent compatibility checks in one call (e.g. ranking candidate products for one device). Each result has its own decision, evidence and receipt.

Call this whenever you're about to recommend, link to or buy a part, accessory, charger, filter, cartridge, lens, battery, memory, mount or adapter for something the person already owns, or they ask "will X work with Y" / "is this compatible". Pass both exact models. Covers anything that has to fit, plug in, mount, run on or charge with something someone owns. Verified catalog (specs checked against cited manufacturer sources): computers & consoles (RAM, M.2 SSDs, USB-C chargers), phones/tablets/watches (USB-C PD, MagSafe, Qi2, watch bands), cameras (lens mounts, adapters, batteries, memory cards), cordless power tools (battery platforms), printers (ink, toner, drums), appliances (refrigerator water and air filters, dryer vents and power cords, installation fit), vacuums (bags, filters, batteries, robot-vacuum parts), air quality (purifier filters, HVAC media, humidifier pads), e-bikes (batteries and chargers by drive system), smart home (Matter, Thread, Zigbee, hubs, bulb bases), PC building (CPU sockets, cooler mounts, memory, clearances, power), cars & EV charging (tire sizes by trim, NACS/J1772/CCS1 and approved adapters, home chargers; a starter set of models) and TVs, consoles & streaming (VESA mounts, weight and screen size, console generations, app platforms). Products outside the catalog in any category are read by AI in terms of the same standards and returned as unverified. Returns a decision (supported / conditional / incompatible / insufficient_evidence), the rules checked, caveats, required accessories, the questions to ask when it can't decide, evidence links and a signed receipt. When the text matches several models it returns each candidate's answer.

Call this when the person names a device or appliance they own and needs a replacement or add-on ("what filter does my fridge take", "which RAM can I add", "what battery fits my drill") before you search shops or guess. Which parts and accessories fit this model? OEM parts first, then verified alternatives: e.g. a refrigerator's water filter, a printer's cartridges, a camera's lenses and batteries, a tool's batteries, or the RAM, SSDs and chargers that fit a computer.

Call this before recommending, buying or selling anything used or secondhand, any baby or kids' product, appliance, heater, battery-powered device or vehicle, or when the person asks if something is safe or recalled. Searches US government recall notices: CPSC for consumer products (appliances, furniture, kids' products, electronics, tools; by brand and model or UPC) and NHTSA for vehicles (by year, make and model). Use before buying, selling or recommending something, especially used or secondhand. Returns decision (recalled / possible_recall / no_recall_found / insufficient_evidence) with each notice's hazard, remedy, which units it applies to and the official link. 'no_recall_found' is not a safety guarantee.

Call this before you tell someone a web page, site or deploy is done or working, or before relying on another agent's "it's live": we open the public page in a separate browser (desktop or mobile) and check what you list. It loads; this text is visible; this link or button leads to this address; no script errors; no broken same-site links. Pass / fail / couldn't-verify per check, with a screenshot and a signed receipt. Read-only: no logins, forms or purchases.

Call this before you recommend, install or rely on an MCP server, after deploying one, or when a connector seems broken: we connect as an MCP client, list its tools, check each has a usable input schema and description, confirm the tools you name exist, and optionally run one sample call on a read-only tool. Pass / fail / couldn't-verify per check with a signed receipt. Never signs in or calls tools that change things.

Call this after building or deploying an MCP server or paid agent API, or when its listings may be out of date: we check it works (as verify_mcp does), then look it up in the official MCP Registry (and whether the listed version is current), Glama, PulseMCP, the x402 Bazaar, npm and PyPI, plus its server card, llms.txt and robots.txt (Smithery, PulseMCP, mcp.so and Claude's directory are listed with links to check by hand). Returns a score, what's missing or stale, an ordered fix plan and a ready-to-publish server.json. Read-only.

Call this after building or changing a paid (x402) endpoint, or when agents' payments fail: we call it unpaid like a paying agent, read the 402, and check the payment request parses (v1 or v2), the token's EIP-712 domain (extra.name/version) matches the contract on-chain (the usual cause of 'invalid signature' payments), network id, amount, payTo, timeout, CORS headers for browser agents, resource URL, description and Bazaar discovery. Pass / warn / fail per check with exact fixes. Never pays.

Call this after launching an MCP server or paid agent API to keep it findable: we run launch_check today and every day for 30 days, and tell your webhook (signed) when something changes: the server stops answering, the registry listing falls behind a release, a directory drops it, robots.txt starts blocking. $3 per 30 days; call again with watch_id to renew. Returns today's full launch check.

Where a watched MCP server stands right now: each directory's status, the latest changes, and when the watch ends. Free.

Call this after a compatibility check, purchase plan, verify_page or verify_mcp result that someone will rely on later (an order next week, an install, a resale, the next agent in a handoff). For product decisions we watch the records, manufacturer sources and US recall notices behind it; for page and MCP checks we re-run the same checks daily. Watches last 30 days and need only the receipt_id. Free during early access.

Call this right after placing an order you checked with check_cart_before_checkout, validate_purchase_plan or check_compatibility: link the receipt_id to the order number, AP2 cart or intent mandate, ACP checkout session, UCP checkout or payment reference. This makes a Decision Record: if the purchase is ever disputed ("the agent bought the wrong thing"), anyone can show what was checked, from which sources, and what changed since. Free.

Call this when an agent purchase is disputed, returned as wrong, or charged back and a Tethered Trust receipt exists for it: we assemble the evidence (the decision, products checked, sources with dates, the linked order or mandate, records and recall notices that changed since, reported outcomes, and signature verification) into one printable page and JSON anyone can verify. For merchants, platforms and buyers. $1.50 per packet.

Call this when a product we said would work didn't (within 60 days of the check): send the receipt_id and what happened. Decided within 10 business days; if our answer or data was wrong, the fee paid for that check is refunded and the data corrected. Free. For a plain fit/no-fit report without a claim, use report_outcome.

Call this right before acting on a watched decision: still_holds, changed (a record it rested on was updated), recall_on_record (a recall notice names one of the products), or expired, with what changed and what isn't monitored.

Will this appliance physically fit this space? Compares dimensions plus typical installation clearances and required connections (power, water, gas).

Check a Tethered Trust receipt: is the signature ours, is it still within valid_until, and have the records behind it changed since? A valid receipt proves what we said and when; it doesn't guarantee the verdict was right.

After a purchase, report whether it actually fit. Free, safe to call, and only the first report per check counts. Reports are weighted by level: self_reported (website), agent_reported (anonymous API/MCP), partner_reported (API key holders) and order_confirmed (order evidence we've confirmed). Only the last two can raise future confidence.