Scheduled Agents
Scheduled agents are workflow agents that run automatically on a cron schedule. They are ideal for recurring tasks like daily reports, weekly compliance checks, or periodic data synchronization.
How Scheduling Works
Section titled “How Scheduling Works”Each scheduled agent is backed by a cron trigger that determines when it runs. Kindo uses standard 5-field cron expressions (minute, hour, day-of-month, month, day-of-week) and supports timezone-aware scheduling.
When you create a scheduled agent, you define:
- A cron expression or natural-language schedule (e.g., “every Monday at 9 AM”)
- A timezone for the schedule to run in
- The workflow steps the agent executes on each run
Scaling Limits and Quotas
Section titled “Scaling Limits and Quotas”Scheduled agents have organization-level quotas to ensure platform stability. These limits apply to the total number of enabled scheduled agents across all users in your organization.
Default Limits
Section titled “Default Limits”| Limit | Default Value | Description |
|---|---|---|
| Max scheduled agents | 5 per organization | Total number of enabled scheduled agents allowed |
| Minimum run interval | 60 minutes | Shortest allowed interval between consecutive runs |
How Quotas Work
Section titled “How Quotas Work”- Quotas are counted at the organization level, not per user.
- Only enabled scheduled agents count toward the quota. Disabling a scheduled agent frees up a slot.
- The quota is checked when creating an enabled schedule or re-enabling a disabled one.
- If your organization has reached its quota, the “Build Agent” button for scheduled agents will be disabled with an explanatory message.
- Your current usage is displayed on the agent creation screen (e.g., “3 / 5 scheduled agents used”).
Requesting Higher Limits
Section titled “Requesting Higher Limits”If your organization needs more scheduled agents or shorter run intervals, contact your Kindo account representative or reach out to support. Limits can be adjusted per organization based on your plan and usage patterns.
Best Practices
Section titled “Best Practices”Choose Appropriate Intervals
Section titled “Choose Appropriate Intervals”- Hourly or less frequent schedules are recommended for most use cases.
- Consider whether your data sources update frequently enough to justify shorter intervals.
- Longer intervals reduce credit consumption and system load.
Design for Idempotency
Section titled “Design for Idempotency”Scheduled agents should produce consistent results when run multiple times. If a run fails or is retried, the agent should handle duplicate data gracefully.
Monitor Agent Runs
Section titled “Monitor Agent Runs”Review your scheduled agent run history regularly in the Agent Monitoring tab. Look for:
- Failed runs that may need investigation
- Runs that take unexpectedly long to complete
- Agents that are no longer needed and can be disabled to free quota
Triage Runs Across Agents
Section titled “Triage Runs Across Agents”In the Agent Monitoring tab, you open each run to see why it failed. To triage runs across your agents with a single request, ask in a Kindo chat with the Platform tool on. For example: “Find my agent runs that failed in the last 24 hours, explain why each one failed, and retry the ones that can be retried.” Runs that a schedule started are not among the ones that can be retried; see the limits below.
Platform includes four tools for this work:
| Tool | What it does |
|---|---|
platform_list_agent_runs | Lists runs across the agents you can access, filtered by agent, status, and creation time |
platform_get_agent_run | Reads a run’s steps, status, and error details |
platform_retry_agent_run_step | Retries a failed step within the same run |
platform_stop_agent_run_turn | Asks a run’s active turn to stop |
An agent with Platform selected in its tools can use them too, and so can an external MCP client through the MCP API. For each tool’s exact behavior, see Inspect and control agent runs in the tool catalog.
The main limits are below; the catalog lists every condition.
- The list includes only runs you have access to. Each request returns at most 100 runs, and the rest continue on the next page.
- Time filters take ISO 8601 timestamps with a timezone, such as
2026-11-05T09:00:00Z. - Retry works only on a failed LLM step in a failed run that was not started by a trigger. A run that a schedule starts counts as started by a trigger, so it cannot be retried this way. To run it again, open the run and select Restart, which starts a new run.
- Retrying needs access to the agent and write access to the run’s conversation, and keeps the agent version the run started on.
- Stop asks the turn to stop, so the run may not halt immediately. Stopping needs write access to the run’s conversation.
Manage Your Quota
Section titled “Manage Your Quota”- Disable unused agents rather than deleting them if you may need them again. Disabled agents don’t count toward your quota.
- Consolidate similar agents where possible. A single agent with multiple steps is often more efficient than several single-step agents.
Creating a Scheduled Agent
Section titled “Creating a Scheduled Agent”-
Navigate to the Agents tab and select Create an Agent.
-
Choose Scheduled as the agent type.
-
Configure your workflow steps (LLM steps, API actions, etc.).
-
Set your cron schedule using the schedule builder or a cron expression.
-
Select the timezone for the schedule.
-
Save and enable the agent.
For a detailed walkthrough, see the Scheduled Vulnerability Report tutorial.
Next Steps
Section titled “Next Steps”- Creating and Managing Agents for general agent setup
- Agents Overview for all agent types
- Scheduled Vulnerability Report for a hands-on tutorial