Risk Scan
$0.01POST /risk-scanScore agent action safety for prompt injection, secret leakage, PII, and payment-risk text before a tool call runs.
Pre-execution checks before agents spend, write, ingest, or hand off work. Boundary Guard gives agents a deterministic checkpoint for risky text, malformed JSON, PII, contact extraction, RAG chunking, tool calls, payment scope, and agent-to-agent handoffs.
Each endpoint is narrow, deterministic, and cheap enough to place directly inside an agent workflow before data is trusted or an action continues.
POST /risk-scanScore agent action safety for prompt injection, secret leakage, PII, and payment-risk text before a tool call runs.
POST /json-extractExtract and repair JSON from messy model output so agents can pass valid structured data downstream.
POST /schema-normalizeValidate and normalize untrusted agent data against JSON Schema before storage, workflow routing, or execution.
POST /pii-redactDetect and redact PII, emails, phones, and secret-like values before agents pay, log, store, train on, or share x402 receipts and payloads.
POST /contact-extractExtract emails, phones, URLs, and domains from scraped pages, inbox text, and lead notes for sales agents.
POST /text-chunkSplit long text into deterministic overlapping chunks with stable hashes for RAG, embeddings, and audit replay.
POST /guardRun a pre-flight guardrail for MCP tools, payments, writes, and workflows; return allow/retry/block plus a Boundary Receipt.
POST /agent-handoffPlan and verify agent-to-agent handoffs with authority, task, payment, constraints, and receipt evidence.
POST /workflow-auditAudit a multi-step agent workflow for risk, spend, missing receipts, and policy evidence before it runs.
POST /receipt-verifyVerify Boundary Guard receipt packages and supplied x402 payment metadata before an agent trusts settlement evidence.
POST /runtime-route-preflightChoose a model/runtime route before an agent spends on inference or bundled APIs; return budget, risk, recommendation, and receipt evidence.
POST /data-enrich-liteNormalize caller-supplied public company, person, URL, or lead hints into a source-linked no-secret enrichment package with receipt evidence.
POST /lead-brief-liteTurn caller-supplied public lead or company facts into an outreach-ready source-linked brief with missing-field flags and receipt evidence.
Most current x402 competitors help sellers charge for APIs. Boundary Guard is the thing a buyer runs after access clears and before the agent trusts the result.
Run Risk Scan or full Guard before MCP tool calls, wallet actions, memory updates, customer messages, or workflow handoffs.
Repair JSON, normalize schema drift, redact PII, extract contacts, and chunk text before storage, retrieval, or CRM enrichment.
Keep a Boundary Receipt or Agent Handoff Receipt that records the policy version, hashes, decision, and operator-readable summary.
Agents can discover the paid host, receive a standard x402 challenge, pay in USDC on Base, and retry with payment proof.
The commercial surface is intentionally easy to understand: one price, one paid host, one free health endpoint, and machine-readable discovery.
per call across 13 premium tools; Workflow Audit and Receipt Verify run at $0.03; Runtime Route Preflight launches at $0.02
Use /healthz for uptime and smoke tests
x402 exact payment in USDC on eip155:8453
The proof is operational, not inflated: the service is live, payment settlement has been verified, and marketplace discovery has refreshed. It is not presented as customer traction.
https://boundary-guard-x402.onrender.comhttps://boundary-guard.vercel.appPOST /workflow-audit is the ninth paid endpoint and settled once on Base mainnet through an internal seed wallet on 2026-06-06POST /receipt-verify is the tenth paid endpoint and settled once on Base mainnet through an internal seed wallet on 2026-06-06POST /runtime-route-preflight is the eleventh paid endpoint for model/runtime spend routing and settled once on Base mainnet through an internal seed wallet on 2026-06-06POST /data-enrich-lite is the twelfth paid endpoint for no-secret public enrichment and settled once on Base mainnet through an internal seed wallet on 2026-06-06POST /lead-brief-lite is the thirteenth paid endpoint for proof-safe sales and research briefs and settled again through the Hermes internal seed wallet on 2026-06-06For x402 API owners, Boundary Guard now packages a launch checklist, wrapper example, marketplace copy, and seed-proof checklist that keeps proof claims inside the evidence.
Use /seller-launch-kit.json to let planning agents reuse the checklist, wrapper example, buyer queries, and proof-safe copy directly.
Use /seller-launch-kit to review the workflow before wrapping a paid endpoint with /guard, /workflow-audit, and /receipt-verify.
Operators and agents can check current category volume, Boundary Guard listing health, revenue proof boundaries, and the next build queue before deciding what to improve.
/market-health.json separates internal seed and indexing proof from customer revenue so listing copy stays honest.
The current build lane is data enrichment and search, with Lead Brief Lite built, seeded, and visible in CDP plus Agentic detail. The next operator task is non-seed buyer conversion.
Each research and operator cycle compares the newest market, product, seed, deploy, revenue, success, and failure evidence against prior runs, then records the next measurable improvement.
/buyer-conversion-guide maps high-volume x402 buyer jobs to the paid endpoint an agent should call first.
/buyer-quickstart gives cold visitors the shortest path from public preview to paid x402 call without implying customer traction.
The factory reads state/last-run.json and state/improvement-history.jsonl so every automation run learns from research, history, successes, failures, deploys, and revenue evidence.
A buyer should not need to decode hashes to understand why a payload was allowed, retried, or blocked.
x402/payment response proves payment settlement when present; Boundary Guard returns receipt evidence for its own request handling.
Every /guard call now returns a plain-English human_summary: what the payload contained, whether Boundary Guard allowed, retried, or blocked it, the receipt hashes, and the suggested operator next step. Paid x402 calls add separate payment proof from the payment response.
{
"human_summary": {
"headline": "Allowed: the request passed Boundary Guard.",
"plain_english": "Boundary Guard accepted the request, normalized count=4 and active=true, and returned a receipt for this exact decision.",
"operator_next_step": "Use the normalized payload and keep the receipt with the downstream action."
}
}
This is the compact endpoint map an agent or tool router needs before deciding what to call.
POST https://boundary-guard-x402.onrender.com/risk-scan POST https://boundary-guard-x402.onrender.com/json-extract POST https://boundary-guard-x402.onrender.com/schema-normalize POST https://boundary-guard-x402.onrender.com/pii-redact POST https://boundary-guard-x402.onrender.com/contact-extract POST https://boundary-guard-x402.onrender.com/text-chunk POST https://boundary-guard-x402.onrender.com/guard POST https://boundary-guard-x402.onrender.com/agent-handoff POST https://boundary-guard-x402.onrender.com/workflow-audit POST https://boundary-guard-x402.onrender.com/receipt-verify POST https://boundary-guard-x402.onrender.com/runtime-route-preflight POST https://boundary-guard-x402.onrender.com/data-enrich-lite POST https://boundary-guard-x402.onrender.com/lead-brief-lite GET https://boundary-guard-x402.onrender.com/healthz
Use /docs for live Swagger exploration, /openapi.json for machine-readable schema discovery, /.well-known/x402 for an x402-native catalog summary, and the paid Render host for x402-protected production and marketplace discovery.
The free preview lets humans verify the response shape before wiring a wallet-capable x402 client to the paid host.
This curl uses the free Vercel preview for a frictionless first check. For paid x402 calls, use the Render host with a wallet-capable x402 client against the same request schema.
curl -sS https://boundary-guard.vercel.app/guard \
-H 'content-type: application/json' \
--data '{
"text": "payload: {\"count\": \"4\", \"active\": \"true\"}",
"schema": {
"type": "object",
"required": ["count", "active"],
"properties": {
"count": {"type": "integer"},
"active": {"type": "boolean"}
}
}
}'