0x96F6eF951840721AdBF46Ac996b59E0235CB985C on eth.
This page builds on the other integration guides — it does not repeat them. For MCP client
setup see MCP Server; for the full endpoint reference and common
workflows see Agent Reference. Scoring methodology is served
live by
GET /v3/framework (public, no key).Prerequisites
- An API key from developers.webacy.co. Base URL:
https://api.webacy.com. All requests send thex-api-keyheader.
Start Here: One Call, Most of the Checklist
For any pegged / RWA token, the MCP toolget_rwa_token_risk or GET /v3/rwa/{address} returns
a per-criterion pass / warn / fail scorecard plus an A+→F grade — answering most of a
structural DD in a single request.
criteria.*.status as one line of the checklist. Here is what the scorecard tells you
about USDY:
Fan Out for Depth
The scorecard is the headline; these endpoints add the underlying detail.Peg Stability and Depeg
peg_value: 1.14), so a price of $1.139 reads as at
peg. The snapshot returns score, tier, a ~12-factor drivers[] decomposition, and
abs_dev_clean (USDY: ~0.0007, ~7 bps off its yield-adjusted peg). Add &hours=168 for up to 7
days of history. This response also carries the liquidity block (liquidity_tier,
total_dex_liquidity_usd, cost_to_move_up_usd / cost_to_move_down_usd, per-DEX markets[])
and any on-chain NAV fields (rwa_nav_usd, rwa_classification, rwa_nav_as_of).
Supply and Mint/Burn Velocity
supply, net_24h_usd, and pressure_vs_30d / pressure_tier to spot supply stress.
The list form GET /rwa/supply sorts the whole universe by mint/burn pressure.
Holder Concentration
top10Index, top30Index (Herfindahl-style), riskBand, holderCount. For deeper
holder forensics (first buyers, snipers, minter) use the get_holder_analysis MCP tool.
Backing, Attestation, and Controls
The v3 scorecard already answers the yes/no control and collateral questions (audited, mint_cap_present, owner_is_multisig, timelock_present, share_price_stable,
mint_burn_anomaly). Deeper reserve composition and third-party attestation freshness
(issuer, provider, attestation type and lag, redemption capacity) are available for covered
assets — reach out to Webacy for access to the reserves surface.
Browse the Universe
GET /rwa filters the RWA + stablecoin universe (by chain, denomination, tags, score);
GET /v3/rwa/grades returns the graded list. In MCP, use list_rwa_tokens.
Coverage Isn’t Binary
Not every asset is fully covered. The list-item fields tell you which state an asset is in:
The list-item classification field is
assetType; has_monitor_data distinguishes “in the
depeg monitor” from merely “asset-type tagged” — check both.
Requesting coverage. There is no self-service coverage endpoint. Send Webacy the symbol +
contract address + chain (and issuer, if niche). RWA classification and monitor coverage are
added by Webacy; coverage lands in stages (monitor / depeg first, then classification + pricing,
then NAV / attestation for permissioned funds). Track it by re-running GET /rwa?q=SYMBOL:
has_monitor_data: true and assetType land first (monitor + classification), then price and
market_cap_usd fill in for full coverage, with NAV / attestation last for permissioned funds.
Read the Provenance, Not Just the Number
Where a value isn’t independently confirmed, the API returnsnull or 404 rather than guessing.
An empty field is a real DD signal — unconfirmed, ask the issuer — not a gap to paper over.
Rule of thumb: on-chain and graded criteria are independently confirmed; a populated
attestation is confirmed as of its date and lag; a null / 404 is unconfirmed. That
combination clears most of a structural checklist in one or two calls, leaving the
judgment-heavy, issuer-specific questions for a human.
Asset-Level vs. Contract-Level
Use asset-level (/rwa/*, /v3/rwa/*) to ask “is this RWA sound?” (peg, NAV, grade,
holders). Use contract-level (/tokens/*, /contracts/*, /addresses/*, and the
get_token_risk / get_address_risk MCP tools) to ask “is this contract or admin address
dangerous?” (bytecode, honeypot, deployer, sanctions). Do both — they complement, not substitute.
MCP Server setup
Connect Claude Code, Cursor, or any MCP client, and the tool reference.
Agent Reference
Full endpoint reference, response shapes, and common workflows.
