Limitations - BlinkOps Documentation

General Limitations

Limit Description	Limit Value	Notes
Maximum size per input	`100KB`	In most cases, the input refers to the output of an ability execution. If this output exceeds the 100KB limit, the agent session will fail and may not respond as expected. Recommendation: To avoid exceeding the size limit, workflow builders should transfer only the most relevant information from the ability execution to the agent input. Use workflow outputs to extract and pass concise, essential data
Maximum size for an agent executed ability (workflow) final output	`100KB`(default) and`300KB` (When ‘Summarize Long Responses’ setting is enabled)	When the ‘Summarize Long Responses’ setting is enabled (Step 4.1 of Building an Agent), the maximum output size for an agent-executed ability is `300 KB`; when the setting is disabled, the maximum is `100 KB`.
Maximum output size for an agent step’s output (final response)	`40KB`
Maximum number of ability executions per agent run	`20`
Maximum timeout for an Agent step	`14 days`	The default time out is set to `30` minutes Paused time doesn’t count toward workflow duration. Visit the Exceptions under More In-Depth Explanation of Timeout Limitations section, for a more detailed breakdown.
Agent Generated Abilities	`5`	If an agent already has 5 agent generated abilities and attempts to generate an additional ability, an error will be displayed indicating that the maximum number of agent generated abilities has been reached.

Agent Depth

To prevent infinite loops, the system limits how many agents can call each other in a chain. This limit is known as the agent call depth. If more than 5 agents are invoked consecutively, either directly via workflow abilities or indirectly through subflows, the system halts execution and raises an error.

Limit Description	Limit Value	Notes
Maximum allowed depth	`5 steps`

Agent Depth Examples

Scenario Description	Agent Call Chain	Allowed?
5 unique agents	Agent A → Agent B → Agent C → Agent D → Agent E	✅ Yes
Same agent called 5 times	Agent A → Agent A → Agent A → Agent A → Agent A	✅ Yes
6 agent invocations (loop pattern)	Agent A → Agent B → Agent A → Agent B → Agent A → Agent B	❌ No

Knowledge - File Upload Limitations

Images included in files uploaded to Knowledge are not supported and will be ignored during processing.

Important to Note: Knowledge sources are not currently supported when using Amazon Bedrock or Azure AI Foundry as the AI provider. Knowledge functionality is available only when using Blink as the AI provider

Limit Description	Limit Value	Notes
Maximum number of documents that can be uploaded	`10`
Maximum file size per document	`50 MB`	Supported document types: `csv` `doc` `docx` `html` `md` `pdf` `txt` `xls` `xlsx`
The total amount of knowledge data processed per task	`0.6 MB`	Agents can query the knowledge multiple times per task.

Rate Limits

Rate limits are applied per tenant and depend on the type of AI request. They protect overall platform availability, so that a spike in requests from one tenant doesn’t affect other tenants.

Limit Description	Interactive	Non-interactive
Applies to	Interactive Agent Mode	Agent Tasks
Maximum requests per minute (per tenant)	`500` requests/minute	`1,600` requests/minute
Maximum agent data usage per minute (per tenant)	`3.58` MB/minute	`23` MB/minute

​General Limitations

​Agent Depth

​Knowledge - File Upload Limitations

​Rate Limits

General Limitations

Agent Depth

Knowledge - File Upload Limitations

Rate Limits