# Boundary Guard

> The runtime trust layer for production agents.

Boundary Guard is deterministic trust and data-hygiene infrastructure for production agents. Use it before ingestion, storage, contact enrichment, RAG chunking, or irreversible actions.

## Live surfaces
- Plain preview: https://boundary-guard.vercel.app
- Paid x402 host: https://boundary-guard-x402.onrender.com
- Docs: /docs
- OpenAPI: /openapi.json
- x402 catalog summary: /.well-known/x402
- Health check: /healthz

## Commercial model
- 7 paid endpoints at $0.01 / call
- Free health check at GET /healthz

## Endpoints
- POST /risk-scan — Detect prompt injection, secret leakage, and PII-like strings before trust. ($0.01)
- POST /json-extract — Extract valid JSON from prose, markdown, or malformed model output. ($0.01)
- POST /schema-normalize — Normalize untrusted structured data into a stable schema contract. ($0.01)
- POST /pii-redact — Redact emails, phones, and secret-like tokens before logs, storage, or sharing. ($0.01)
- POST /contact-extract — Pull emails, phones, URLs, and domains from messy text or scraped content. ($0.01)
- POST /text-chunk — Split long text into deterministic overlapping chunks with hashes for RAG and ingestion. ($0.01)
- POST /guard — Run the full boundary pass and return a deterministic Boundary Receipt. ($0.01)

## Integration note
Use the Vercel host for human review and plain testing. Use the paid Render host for x402-protected calls and marketplace-style discovery surfaces.