Agent Verification Utility
by agent-economy.kgninja.dev · remote · no sign-in to list tools · v0.4.0 · published by dev.kgninja.agent-economy
Deterministic JSON checks with signed evidence and x402-paid execution
5 tools: 1 do something that can't be undone, 2 change things and 2 only read.
Scanned Oct 21, 2026
- Deletes or cancels0Claude Code: deny
- Can't be undone1Claude Code: ask: see these tools
- Changes things2Claude Code: ask: see these tools
- Only reads2Claude Code: allow: see these tools
No check matched anything in the tool definitions. No finding is not a safety guarantee. How the checks work.
claude mcp add --transport http agent-verification-utility https://agent-economy.kgninja.dev/mcpTools
Open a tool for its reasons and full description. Descriptions are the publisher's own words, shown as they are.
5 tools
Can't be undone 1Claude Code: ask
prepare_verify_evidence_purchasePrepare an HTTP purchase request for deterministic JSON verification after a successful free precheck; a prior quote tool call is optional.No findings
Why it is in this group
- The name "prepare_verify_evidence_purchase" says "purchase": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Prepare an HTTP purchase request for deterministic JSON verification after a successful free precheck; a prior quote tool call is optional. After a successful free precheck at POST /validate-request, supply the unchanged intent and its precheck_receipt.receipt_digest. Creates or reuses a persisted quote and records observability events; no payment, signing or paid verification is performed. The bound quote ties request/evidence digests, precheck digest, spend policy, quoted amount, network, asset, recipient and expiry together. Server availability, fresh cost basis and conservative contribution-margin thresholds must pass. Returns JSON text and structuredContent containing quote, request (POST /verify-evidence, content-type and X-Quote-ID headers, unchanged intent body), expected_first_status=402 and expected_payment_header=PAYMENT-REQUIRED. It does not send this request. Application failures return isError=true with JSON error.code, message, retryable and optional details (for example PRECHECK_STALE, PRICE_CAP_EXCEEDED, IDEMPOTENCY_KEY_CONFLICT, IDEMPOTENCY_KEY_EXPIRED or UNPROFITABLE_TRANSACTION); invalid argument shapes are rejected by MCP schema validation. Re-run precheck after request/policy changes; do not pay on refusal. Next send the returned request to obtain a challenge, check binding, amount, network, asset, payTo and expiry against local policy, then authorize payment separately. Use quote_verify_evidence when only quote terms are needed.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
Changes things 2Claude Code: ask
quote_verify_evidenceObtain a free bound quote for deterministic checks on supplied JSON before deciding to purchase.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Description, as the server wrote it Obtain a free bound quote for deterministic checks on supplied JSON before deciding to purchase. After a successful free precheck at POST /validate-request, supply the unchanged intent and its precheck_receipt.receipt_digest. Creates or reuses a persisted quote and records observability events; no payment, signing or paid verification is performed. The bound quote ties request/evidence digests, precheck digest, spend policy, quoted amount, network, asset, recipient and expiry together. Server availability, fresh cost basis and conservative contribution-margin thresholds must pass. Returns the quote as JSON text and structuredContent: quote_id, request_hash, price (atomic USDC, 6 decimals), expires_at, economic_guard, purchase instructions and paid_verification_binding in state quoted, not a paid receipt. Application failures return isError=true with JSON error.code, message, retryable and optional details (for example PRECHECK_STALE, PRICE_CAP_EXCEEDED, IDEMPOTENCY_KEY_CONFLICT, IDEMPOTENCY_KEY_EXPIRED or UNPROFITABLE_TRANSACTION); invalid argument shapes are rejected by MCP schema validation. Re-run precheck after request/policy changes; do not pay on refusal. Next use prepare_verify_evidence_purchase with the same key and intent for the HTTP purchase request, or explicitly authorize the separate paid verify_evidence flow.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
verify_evidenceRun deterministic checks and return Ed25519-signed evidence after one explicitly authorized x402 payment.No findings
Why it is in this group
- The description's first verb is "run": a word for a change.
Description, as the server wrote it Run deterministic checks and return Ed25519-signed evidence after one explicitly authorized x402 payment. A free precheck receipt and a matching bound payment requirement are required before signing.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
Only reads 2Claude Code: allow
describe_verify_evidenceDescribe when deterministic verification is useful, when it is not, free request validation, recipes, limits, and the explicit paid MCP path.No findings
Why it is in this group
- The name "describe_verify_evidence" says "describe": a word for reading.
Description, as the server wrote it Describe when deterministic verification is useful, when it is not, free request validation, recipes, limits, and the explicit paid MCP path.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
get_revenue_goal_statusRead the active revenue goal, measured funnel, contribution margin, and the next bounded improvement recommendation.No findings
Why it is in this group
- The name "get_revenue_goal_status" says "get": a word for reading.
Description, as the server wrote it Read the active revenue goal, measured funnel, contribution margin, and the next bounded improvement recommendation. This tool never changes price, spends funds, or sends messages.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.
Install
Pinned to version 0.4.0. Values in angle brackets are yours to fill in.
claude mcp add --transport http agent-verification-utility https://agent-economy.kgninja.dev/mcpRun in a terminal.
Claude Code permissions
Allow the reads, ask before a change, deny the deletes. Paste into .claude/settings.json.
{
"permissions": {
"allow": [
"mcp__agent-verification-utility__describe_verify_evidence",
"mcp__agent-verification-utility__get_revenue_goal_status"
],
"ask": [
"mcp__agent-verification-utility__quote_verify_evidence",
"mcp__agent-verification-utility__prepare_verify_evidence_purchase",
"mcp__agent-verification-utility__verify_evidence"
],
"deny": []
}
}Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. It assumes the server is installed as agent-verification-utility. Review it before you use it.