Supercharge your AI agents with data from the web and beyond. Building the library for superintelligence. 🔥
TypeScript aiai-agentsai-crawlerai-scrapingai-searchcrawlerdata-extractionhtml-to-markdownllmmarkdownscraperscrapingweb-crawlerweb-dataweb-data-extractionweb-scraperweb-scrapingweb-searchwebscraping
import type { Request } from "express";
import { config } from "../config";
import type { AuthCreditUsageChunk } from "../controllers/v1/types";
import { getRoutePattern } from "./http-metrics";
import { logger } from "./logger";
export type AuthDenialReason =
| "missing_credentials"
| "malformed_authorization"
| "malformed_key"
| "unknown_key"
| "invalid_mcp_credential"
| "invalid_oauth_token"
| "oauth_purpose_mismatch"
| "oauth_team_mismatch"
| "invalid_vercel_oidc_token"
| "vercel_oidc_team_mismatch"
| "team_banned"
| "ip_restricted"
| "endpoint_restricted"
| "keyless_ip_ineligible"
| "keyless_ip_suspicious"
| "keyless_limiter_unavailable"
| "agent_key_blocked"
| "agent_key_verification_expired";
// Admin routes carry BULL_AUTH_KEY as a path segment.
function routeOf(req: Request): string | undefined {
if (typeof req.path !== "string") return undefined;
return `${req.baseUrl ?? ""}${getRoutePattern(req)}`
.split("/")
.map(segment =>
config.BULL_AUTH_KEY && segment === config.BULL_AUTH_KEY
? ":bullAuthKey"
: segment,
)
.join("/");
}
/**
* Writes the single `auth/denied` line for a request refused access. Pass `key`
* only when it came from a resolved API key: preview and keyless chunks carry
* the client IP in their team id. Never log the credential or the client IP.
*/
export function logAuthDenied(
req: Request,
status: number,
reason: AuthDenialReason,
key?: Pick<AuthCreditUsageChunk, "team_id" | "api_key_id">,
): void {
logger.warn("Request denied", {
canonicalLog: "auth/denied",
reason,
status,
method: req.method,
route: routeOf(req),
teamId: key?.team_id,
apiKeyId: key?.api_key_id,
});
}