Agent Input Guard
Prompt injection and tool abuse heuristic scanner with x402 USDC payment on Base. Verified route: POST /scan at 0.002 USDC, taking {text} and returning a risk level, score, finding count, matched patterns and a recommendation. Verified with a non advertised credential exfiltration prompt, where it returned a structured medium risk result quoting the matching substring from the submitted text. Heuristic only: false positives and misses are possible. Base only.
1 route settlement-verified · $0.002 USDC/request · 1 endpoint · Tools · base network · x402 protocol
Verified routes
POST /scan· $0.002 · 2026-09-15 · receipt
Other routes are unverified.
Networks: Base (settlement-verified)
Discovery is free. Agents search the catalogue here. Payment happens on the listing 402, not on PayAPI.
url https://payapi.market/mcpListing facts
- Provider
- Project Zero Dollars
- Hosted at
project-zero-agent-api.jordy-prods.workers.dev- Category
- Tools
- Price
- $0.002 USDC per request
- Endpoints
- 1
- Payment networks
- Base
- Settlement-verified
- 1 route, last checked 2026-09-15
- Listed on PayAPI Market
- 2026-09-14
Try Agent Input Guard
Call a verified route without paying first. The server should answer HTTP 402 Payment Required with the price, the USDC asset, the network and the wallet to pay. An x402 client pays that amount on Base and retries with the payment header.
curl -i -X POST "https://project-zero-agent-api.jordy-prods.workers.dev/scan"About this API
Agent Input Guard is listed in the Tools category with pricing at $0.002 USDC/request. One route has been settlement-verified by PayAPI Market on 2026-09-15. View receipt on Basescan.
Payments settle on base in USDC via the x402 protocol. AI agents discover this API through the PayAPI Market MCP server and pay per request without API keys or accounts.
This API exposes 1 endpoint.
Questions about this API
What does Agent Input Guard do?
Prompt injection and tool abuse heuristic scanner with x402 USDC payment on Base. Verified route: POST /scan at 0.002 USDC, taking {text} and returning a risk level, score, finding count, matched patterns and a recommendation. Verified with a non advertised credential exfiltration prompt, where it returned a structured medium risk result quoting the matching substring from the submitted text. Heuristic only: false positives and misses are possible. Base only. It is provided by Project Zero Dollars.
How much does Agent Input Guard cost?
Agent Input Guard is priced at $0.002 USDC per request, paid per call with no subscription. PayAPI Market paid 0.002 USDC for one call to /scan when it verified the listing. The exact amount for each call is in the HTTP 402 response before any money moves.
Do I need an API key or account to use Agent Input Guard?
No. Agent Input Guard uses the x402 protocol, so an agent pays per request in USDC on Base and gets the response, with no API key, account or signup.
Has Agent Input Guard been verified?
Yes, 1 route. PayAPI Market paid for a real call from its own wallet, received real data back and recorded the USDC settlement on-chain, most recently on 2026-09-15.
How does an AI agent find Agent Input Guard?
Through the free PayAPI Market MCP server at https://payapi.market/mcp, using search_apis or get_api, or over plain HTTP at https://payapi.market/agent/search. Discovery is free; the agent pays the provider directly on the listing's own 402.
More Tools APIs · Browse the marketplace · List your own API
Related APIs
- AgentSiteAudit · 1 route verified · Website audit API for AI agents: 69 checks across security, SEO, TLS, performance, accessibility, DNS and agent readiness. One JSON report with exact fixes, plus Verify.
- StableDomains Domain Valuation · 1 route verified · Domain valuation for AI agents, paid per call in USDC over x402. Part of StableDomains, an agent-first registrar with a 37-tool MCP server.
- Agent Extraction API: Messy text + JSON Schema to clean JSON · 1 route verified · Messy text + JSON Schema in, clean JSON out. Uses a small LLM (Cloudflare Workers AI, Llama 3.1 8B, JSON mode) when available, else a deterministic key:value/regex extractor; the result is type-checked against your schema and missing fields are null. Limits: text <= 6,000 chars, schema <= 30 top-level properties (flat objects, arrays of scalars); the model can misread ambiguous text, so check important values. Params: text, schema (JSON string), mode=auto|heuristic. GET query or POST JSON body.
- Agent Extraction API: Email check without SMTP (syntax, MX, disposable) · 1 route verified · Email address check WITHOUT sending mail or SMTP probing: RFC-style syntax check, domain MX/A records via DNS-over-HTTPS, disposable-domain flag (CC0 blocklist, ~9,900 domains), role-address flag (info@, support@...), free-mail flag. Verdict: deliverable_domain | risky | invalid. Cannot confirm that the mailbox itself exists. Addresses are not stored or logged.
- Agent Extraction API: Web page to clean Markdown · 1 route verified · Convert a public web page to clean Markdown (title, headings, paragraphs, lists, links, code), scripts/nav/footer removed, max ~60k chars. robots.txt respected.
- Fresh402 · 1 route verified · Bounded Web and finalized Base observations for agents over REST and MCP: Check, Extract, Smart Diff, OpenAPI Contract Diff, Contract Watch, Wallet Watch and Token Facts, plus free Register. Listed price is Check ($0.005); other per-call prices and one-time prepaid packages are documented at https://fresh402.com/pricing. Native USDC on Base; x402 v2 and native MPP. Machine trial uses an unfunded EOA ownership signature, no CAPTCHA or payment: up to 2 Checks and one of each other tool, subject to finite shared best-effort capacity, seven-day maximum and current operating expiry. Setup and limitations: https://fresh402.com/docs#machine-trial. Source SDK: https://github.com/kirillradchenko96/fresh402/tree/efficiency-sdk-v0.3.5/examples/efficiency.