Give your agent a controlled API boundary.

1. Connect your editor

https://api.xguardgate.com/mcp

Transport: Streamable HTTP. Download your configuration and merge it into your existing configuration file; preserve your other servers.

• Cursor JSON → .cursor/mcp.json
• VS Code JSON → .vscode/mcp.json
• Claude Code JSON → project .mcp.json
• Codex TOML → your Codex configuration

Or connect Claude Code from your terminal:

claude mcp add --transport http xguard https://api.xguardgate.com/mcp

A connected editor can discover tools. Connection alone does not supply a funded wallet, authorize payment, or provision a provider account.

2. Inspect without paying

Ask your agent: “List XGuard capabilities and prices without executing a paid tool.”

curl https://api.xguardgate.com/v1/capabilities

Live prices · OpenAPI · A2A agent card

3. Make your first payment-request call

curl -i https://api.xguardgate.com/v1/tools/web.fetch -H 'content-type: application/json' -d '{"url":"https://example.com/"}'

Expected response: HTTP 402 with a signed quote and payment instructions. This first call does not pay or contact the target. To execute, a funded compatible x402 v2 payer must sign the challenge and retry the identical request. Check the signed price before approving payment.

4. Use credential-backed API actions

For Secretless Egress, the operator supplies an authorized provider credential and XGuard Usage Credits, then delegates a short-lived capability scoped to host, path, method and budget. Provider charges remain separate. Keep reusable credentials outside the agent context.

Use one stable Idempotency-Key per write operation. Completed retries replay the stored signed outcome; ambiguous in-flight writes fail closed and are not automatically retried.

Execution contract and limits · SDK examples and installation · Credits and pricing