Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure
TypeScript
import type { Request, Response, Router } from 'express'
import { ipKeyGenerator } from 'express-rate-limit'
import { config } from '../../config.js'
import { type UiAuthedRequest, requireAuthForControlUI } from '../../middleware/controlUIAuth.js'
import { rateLimitMiddleware } from '../../middleware/rateLimitMiddleware.js'
import { authenticateAdminSession } from '../../services/adminSessionAuth.js'
import { CONTROL_UI_ADMIN_SESSION_COOKIE, readCookie } from '../../utils/auth/sessionCookies.js'
import { parseRequestedEntityChangeCursor, streamEntityChanges } from '../entityChangeStream.js'
export function registerGfsEntityChangeRoutes(router: Router): void {
const ipRateLimit = rateLimitMiddleware({
bucketType: 'gfs_operator_entity_changes_ip',
maxPerMinute: config.adminPublicTokenIpRlPerMin,
getBucketKey: req =>
`gfs-operator-entity-changes:ip:${ipKeyGenerator(req.ip || req.socket.remoteAddress || 'unknown')}`,
onBackendUnavailable: 'closed',
})
const operatorRateLimit = rateLimitMiddleware({
bucketType: 'gfs_operator_entity_changes',
maxPerMinute: config.adminPublicTokenRlPerMin,
getBucketKey: req =>
`gfs-operator-entity-changes:operator:${(req as UiAuthedRequest).adminAuth?.sub ?? 'unknown'}`,
onBackendUnavailable: 'closed',
})
router.get(
'/gfs/entity-changes/stream',
ipRateLimit,
requireAuthForControlUI,
operatorRateLimit,
(req: Request, res: Response) => {
const adminReq = req as UiAuthedRequest
const cursor = parseRequestedEntityChangeCursor(req)
if (cursor === false) {
res.status(400).json({ error: 'Invalid entity change cursor' })
return
}
const token = readCookie(req, CONTROL_UI_ADMIN_SESSION_COOKIE)
streamEntityChanges(
req,
res,
cursor,
async () => {
const current = await authenticateAdminSession(token)
return Boolean(current && current.sub === adminReq.adminAuth?.sub)
},
'operator',
`operator:${adminReq.adminAuth?.sub ?? 'unknown'}`
)
}
)
}