[ AUTOMATIONS ]

Cron Jobs, Secret Rotation & AI Workflows

Schedule recurring tasks, rotate secrets automatically, and run AI-driven workflows on a cron. No infrastructure to manage.

Get started freeRead the docs

[ 01 ] KEY FEATURES

Everything you need for agent automation

From simple cron schedules to complex AI-driven workflows — all managed through a single API and dashboard.

14 Step Types

HTTP requests, AI generation, memory operations, notifications, secret rotation, transaction signing, conditional branching, and more.

4 Trigger Types

Cron schedules, webhook URLs, event-driven (secret/policy lifecycle), and manual API triggers.

Conditional Logic

skip_if and run_if expressions on any step. Compare outputs, check status codes, branch on template variables.

AI Generation Steps

LLM-powered steps with configurable provider, model, and max tokens. Routes through Shroud when enabled.

10 Built-in Presets

DCA buys, health checks, credential rotation, competitor monitoring, lead nurture, and more — production-ready templates.

Template Variables

Pass data between steps with {{steps.name.output}} syntax. Webhook payloads and event triggers inject context automatically.

Human Approval Gates

Pause a run for human review via approval_request steps. Configurable risk tiers and action summaries.

NL-to-Workflow Assistant

Describe what you want in plain English. The assist endpoint drafts a workflow_spec you can review and deploy.

Multi-Channel Notifications

Send alerts via webhook, Slack, or email from any step. SSRF-protected URLs and template variable support.

Execution History

Full run history with per-step output, status, duration, and error details. Cancel running or awaiting-approval runs.

Secret Rotation Steps

Server-side rotate_generate steps with configurable length and charset. Zero-downtime rotation with version history.

Production Guardrails

5-minute run timeout, max 5 concurrent runs per agent, tier-based monthly quotas, and budget controls for AI steps.

[ 02 ] HOW IT WORKS

Four steps to production automation

Define, configure, protect, and monitor. No infrastructure to provision, no cron daemons to babysit.

how it works
1

Define trigger

Cron expression, webhook URL, event listener, or manual API call. Pick the trigger that fits your use case.

2

Build workflow_spec

Chain 14 step types — HTTP, AI generation, memory, notifications, transactions — with template variables and conditional logic.

3

Set guardrails

Execution time limits, concurrency caps, approval gates, and budget controls for AI steps.

4

Monitor runs

Full execution history with per-step output, status, duration, and cancel support for running workflows.

SDK example
const automation = await client.automations.create({
  name: "Health Check & Alert",
  agent_id: agentId,
  trigger_type: "cron",
  cron_expr: "*/5 * * * *",
  workflow_spec: [
    {
      name: "check",
      type: "http",
      params: { url: "https://api.example.com/health", method: "GET" }
    },
    {
      type: "ai_generate",
      skip_if: "{{steps.check.http_status}}  == 200",
      params: {
        prompt: "Analyze failure: {{steps.check.output}}",
        max_tokens: 256
      }
    },
    {
      type: "notify",
      skip_if: "{{steps.check.http_status}}  == 200",
      params: {
        channel: "webhook",
        url: process.env.SLACK_WEBHOOK,
        text: "Health check failed: {{steps.1.output}}"
      }
    }
  ]
});

Sub-second scheduling precision. Automations fire within milliseconds of their scheduled time. No cold starts.

Sub-second

scheduling precision

0

trigger types

Full

audit trail

[ 03 ] USE CASES

Built for real-world agent operations

Whether you're rotating secrets weekly, reacting to webhooks, or running AI workflows on a schedule — automations handle it.

Credential rotation

Rotate database passwords, API keys, and tokens on a schedule. New credentials are versioned — rollback is instant.

Event-driven pipelines

Trigger workflows from GitHub pushes, Stripe events, or custom webhooks. Chain multiple actions with conditional logic.

AI-powered decisions

Let an LLM analyze logs, classify alerts, or draft responses as part of a multi-step workflow. Budget-capped per run.

Health checks

Ping endpoints on a schedule. Alert on failure. Auto-remediate with retry or failover actions.

[ 04 ] GUARDRAILS

Production-grade safety by default

Every automation runs with configurable guardrails. No uncapped spend, no runaway processes, no silent failures.

Max concurrent runs

Prevent resource exhaustion by capping how many instances of an automation run at once.

Execution time limit

Hard timeout per run. Long-running workflows fail gracefully instead of hanging forever.

Failure thresholds

Auto-disable after N consecutive failures. Get alerted and fix before re-enabling.

Budget controls

Cap AI step spend per run and per day. Prevents runaway LLM costs in multi-step workflows.

Retry policy

Configurable backoff, max retries, and retry conditions. Transient failures resolve themselves.

Run quotas

Monthly run limits per tier. Predictable billing with clear overage controls.

[ 05 ] GET STARTED

Automate your agent operations today

Included on every plan. Free tier: 2 automations, 100 runs/month.

Get started freeRead the docs

Included on every plan. Free tier: 2 automations, 100 runs/month.