# Overwing

> Let agents operate your legacy systems, safely. Tower clears each write an agent makes on a system of record (run, review or refuse, with a signed receipt), Guardrails checks every message before it ships, and alongside them: open data on AI agents, a check of whether a product is reachable by agents, and a check of a transaction or an x402 payment before an agent signs it. Every operation works over plain HTTPS and JSON, with no CAPTCHA and no browser-only step.

## Products

- **Guardrails.** Send any text, usually a model's output, and get a verdict (pass, fail or review), one recommended action (block, redact, review or allow), and per-rule results with calibrated confidence, usually in under 500 ms. https://overwing.ai/
- **Preflight.** Should the agent sign this? One call simulates an unsigned Solana transaction, works out what it would take from your wallet, and answers allow or refuse against limits you set. Every verdict is published. One evaluation a check; $0.01 over x402 without an account. https://overwing.ai/preflight
- **Atlas.** A registry of AI crawlers, fetchers and browser agents, a lookup that says what a User-Agent claims to be and whether the claim can be trusted, and first-hand scans of whole sectors. https://overwing.ai/atlas
- **Beacon.** Is your product reachable by agents? One lookup of a site that says whether agents can find it, read it and use it, with the changes worth making. Free with an account; $1 over x402 without one. https://overwing.ai/beacon
- **Tower.** A decision, permission and receipt layer for agents that operate legacy business systems. Each write is executed, sent to a person, or refused, and every step gets a signed receipt. https://overwing.ai/products/tower

## Two ways to start

- **The send gate.** Guardrails placed between an agent's draft and the send: one call answers allow, redact, review or block, and knows who the recipient is. https://overwing.ai/gate
- **Agent-ready, done for you.** Overwing makes a product findable, readable and usable by agents for a fixed price, from $1,500, delivered as pull requests. https://overwing.ai/services

## Guardrails pricing

| Plan | Price per month | Evaluations per day |
|---|---|---|
| Free | free | 250 |
| Starter | $29 | 5,000 |
| Growth | $99 | 25,000 |
| Scale | $299 | 100,000 |
| Prepaid top-up, card | from $5 for 2,000 | spent after the daily allowance, never expire |
| Pay per request | $0.002 each, in USDC over x402 | no account needed |

Everything Overwing sells, on one page: https://overwing.ai/pricing. Prices and limits are served live at https://overwing.ai/api/v1/plans. Use that, not this page, when a number matters.

## Start

```
curl -X POST https://overwing.ai/api/v1/signup -H "Content-Type: application/json" \
  -d '{"email":"you@example.com","password":"at-least-12-chars"}'
curl -X POST https://overwing.ai/api/v1/evaluate -H "Authorization: Bearer ow_live_..." -H "Content-Type: application/json" \
  -d '{"input":"text to check","rule_set":"content-safety"}'
```

## For agents and tools

- Guide: https://overwing.ai/llms.txt
- API reference: https://overwing.ai/docs.md (markdown), https://overwing.ai/api/v1/openapi.json (OpenAPI 3.1)
- MCP server: hosted at https://overwing.ai/mcp (no install), or `npx -y overwing-mcp`
- A2A: https://overwing.ai/a2a
- SDKs: `npm install overwing`, `pip install overwing`
- OpenClaw: plugin `overwing-openclaw` (every outbound message checked), skill `overwing` on ClawHub
- Discovery: https://overwing.ai/.well-known/mcp.json, https://overwing.ai/.well-known/agent-card.json