Skip to main content
A tool computes over input you send. A resource is static information bundled into the server. WalletWall MCP exposes exactly eleven resources so a client can understand its contract without reading the source repository.

Build valid evidence

Schemas, claim contracts, authority policies and the WWEP draft.

Understand the bundled world

Stablecoin deployments, chain metadata and the versions actually running.

Learn by example

Four complete synthetic inputs with the exact tool calls and expected outcomes.

Tools vs resources

The resource set is fixed when the server is built. There are no resource templates, subscriptions or prompts, and nothing is loaded remotely. Each resource’s SHA-256 is pinned in security/tool-manifest.json.

The eleven resources

Build valid evidence

Understand the bundled world

Learn by example

All four examples are synthetic. Their shape may be reused; their evidence may not. Never present an example as evidence about a real wallet, balance or chain state.

Contract and specification

The bundle schema accepted by inspect_evidence, check_invariants, trace_provenance and assess_vault_readiness: wwepVersion ("0.1"), sources, observations, assertions, derivations, and optional conflicts, supersessions, registry version and rules version. Bundle records are closed; a claim’s subject and value and an observation’s payload are bounded open maps.
The per-claim contract generated from the bundled policies and evaluator rules: required subject and value keys, accepted origins and source kinds, observation linkage, registry/derivation relationships and the diagnostics attached to shape failures.
The authority rules for the five claim types. Authority is defined per claim type; there is no global ranking of source kinds. See Evidence Model (WWEP).
The WalletWall-defined WWEP 0.1 draft: data model, states, invariants, canonicalization, evaluation procedure and known limitations. It is not an industry standard.

Reference data

walletwall://registry/stablecoins contains five deployments, each verified once against an issuer or protocol publication and a finalized decimals() reading on 2026-10-24. walletwall://registry/chains contains Ethereum Mainnet, Base and Arbitrum One with a descriptive finality model for each. The engine does not independently verify finality; it evaluates the finality each observation declares.
The registry is a point-in-time snapshot, not live chain data. Registry-derived results carry REGISTRY_VERIFIED_POINT_IN_TIME; the snapshot says nothing about later contract changes, safety or solvency.

Versions and digests

Read walletwall://conformance/version when you need to know what the running server actually contains. At the time of writing it reports:
status is the engine’s own pre-release label. These digests identify the registry, policies and specification used for a result; the resource does not carry a digest of the engine bundle itself.
1

Read the claim contract

Start with walletwall://schemas/claims and the synthetic example closest to the claim you need.
2

Assemble your own evidence

Reuse the example’s shape, not its evidence. Your measurements remain your inputs.
3

Inspect before deriving readiness

Call inspect_evidence. If the result contains shape diagnostics, fix the bundle and treat the corrected bundle as new input.
4

Assess vault readiness only when it applies

assess_vault_readiness is an evidence-only classification over the required identity and balance assertions. It does not configure, authorize or execute a vault.

Unknown resource URIs

URIs are exact strings. Reading anything outside the eleven listed above returns a JSON-RPC error with Unknown resource. Asking for resource templates or prompts returns Method not found, because the server offers neither.

Understand WWEP

See how the resources become evidence states and assessments.

See WallPrint

Learn how a completed assessment can become a deliberately published disclosure.