How Scalekix operates under the hood
A credential and authorization layer connecting autonomous agents to your infrastructure.
One gateway that recognizes every agent
Each agent request travels through Scalekix, which verifies the agent identity and applies the declared permission scope before sending it to your target systems.
Verified identities for agents
Cryptographic credential for each agent
Each registered agent receives a signed credential bundle with its own skx:// URI. Agents never share credentials, even when they perform the same role or task type.
Adjustable credential lifetime
Choose a credential TTL for each agent, from hours through months. The SDK refreshes credentials automatically before expiration, preventing mid-task failures caused by stale credentials.
Immediate revocation without interruption
Cancel one agent credential with a single API call. There are no shared tokens to rotate, and no other agents or humans are impacted. The gateway receives the revocation within seconds.
Stable agent identity
An agent's skx:// ID remains unchanged through credential rotations, allowing the audit log to associate each action with a stable identity instead of a temporary token string.
Fine-grained permission scopes
Scalekix uses resource-level scope syntax to specify precisely what each agent can do. Scopes use a resource:action pattern: db:read, api:call, files:list, storage:write.
Set the permitted scope collection when you register an agent. Scalekix denies requests outside that declared set, including implied wildcards. Wildcard access must be declared explicitly.
- Unlimited scope rules for each agent on Builder and Platform tiers
- Registration-time schema linting identifies typos and unrecognized resources
- Inherited scopes for agent families using a shared foundational policy
- Custom permission namespaces available on Platform tier
// Agent registration payload
{
"name": "research-agent-3",
"scopes": [
"db:read",
"api:call",
"files:list"
],
"ttl_hours": 720,
"tags": {
"team": "data-platform",
"env": "production"
}
}Audit logs that reveal tampering
Structured JSON activity events
Each agent action creates a structured JSON event containing the agent ID, scope used, target endpoint, request timestamp, and result status. Dashboard filters include agent, time range, and scope.
Timeline for each agent
Every agent has a dedicated activity timeline, separate from human activity logs. Compliance and security teams can inspect agent behavior without human actions creating noise.
SIEM and webhook export
Send audit events to your SIEM in real time through a webhook, or retrieve structured JSON exports on a schedule. Works with common log-ingestion pipelines used by security teams.
Retention settings by tier
Pilot includes 30-day retention, Builder includes 12-month retention, and Platform includes unlimited retention. Export logs to your own storage before expiration when you need a longer local archive.
Connects to the tools you already ship
Python SDK, Node.js SDK, and REST API. Add Scalekix to your agent runtime in an afternoon. The quickstart guide includes complete documentation.