Supply

Tools and capabilities

A capability is something an agent can buy: a priced, schema-checked operation with known payment terms and known proof.

The capability model

Field	What it tells the agent
`id, version, provider`	Exactly what is being bought and from whom.
`inputSchema, outputSchema`	Strict JSON Schemas. Unknown input properties are rejected.
`pricingModel`	Scheme, maximum amount, unit price for metered routes, billing rule and the pricing formula version.
`supportedNetworks, supportedAssets`	Where and in what it can be paid.
`receiptCapabilities, evidenceCapabilities`	What the receipt commits to, and what evidence the result carries (for example block_pinned or source_hash).
`riskMetadata`	Side-effect class, data sensitivity, and whether output is third-party text.
`health`	Latest provider health observation.

Finding and quoting

GET /v1/catalog lists every capability with its schemas and prices. Free.
POST /v1/resolve plans a purchase for a need under constraints. See Resolve.
POST /v1/find is the simpler search used by older clients: deterministic filters, then a documented score.
POST /v1/quote returns a quote bound to the tool version, its definition hash, a hash of your input, the price, network and an expiry.

A call that presents a quote must match it. A different input, an unknown quote or a changed tool definition returns 409 and nothing is charged. The receipt then commits to the quote that was paid for, so an overcharge is provable.

Billing

Billing is success-only. Invalid input returns 400 and a failing upstream returns 502; neither is charged. Metered capabilities report units, and the charge is the formula applied to those units, capped at the signed ceiling.

First-party Robinhood Chain tools

Tollex ships 34 first-party capabilities for Robinhood Chain: chain reads, wallets, USDG, x402 inspection, Stock Token pricing, merchant checks and receipts. Prices range from 0.0003 to 0.005 USDG per call. See Pricing for the full list.

Category	Capabilities
chain	`chain_get_block`, `chain_get_transaction`, `chain_get_receipt`, `chain_get_logs`, `chain_contract_read`, `chain_simulate`, `chain_get_code`, `chain_estimate_gas`
x402	`x402_inspect_endpoint`, `x402_get_requirements`, `x402_validate_requirements`, `x402_simulate_payment`, `x402_facilitator_status`, `x402_reconcile_operation`, `x402_verify_receipt`
wallet	`wallet_native_balance`, `wallet_token_balance`, `wallet_allowance`, `wallet_authorization_state`, `wallet_activity`
stock-token	`stock_tokens_list`, `stock_token_metadata`, `stock_token_quote`, `stock_token_halt_status`, `stock_token_corporate_actions`
usdg	`usdg_metadata`, `usdg_domain`, `usdg_capabilities`, `usdg_authorization_state`
merchant	`merchant_inspect`, `merchant_health`, `merchant_payment_options`
receipt	`receipt_get`, `receipt_verify`

Output is data

Some capabilities return text written by third parties. Their results are marked untrustedOutput, and the MCP server labels them for the model. Nothing in a tool result can change policy, payment or routing.