Max number of concurrent agents in ParallelAgent

[VenkataSakethDakuri]

Is there any way to set the maximum number of concurrent parallel agents in ParallelAgent to avoid rate limits?

[adk-bot]

Response from ADK Answering Agent (experimental, answer may be inaccurate)

TLDR: The ParallelAgent in ADK is designed to run all of its sub-agents concurrently and does not currently offer a built-in configuration parameter to limit the maximum number of parallel executions.

---

Hello! Based on the ADK documentation, the ParallelAgent is a workflow agent that executes all of its provided sub-agents concurrently to maximize performance for independent tasks. [2] Its main purpose is to speed up workflows by running tasks in parallel, and the current implementation is designed to run all of them at once. [1, 2]

The code examples for ParallelAgent show it being initialized with a list of sub-agents, all of which are executed in parallel when the agent runs. [1]

# Conceptual Code: Parallel Information Gathering
from google.adk.agents import ParallelAgent, LlmAgent

fetch_api1 = LlmAgent(name="API1Fetcher", instruction="Fetch data from API 1.")
fetch_api2 = LlmAgent(name="API2Fetcher", instruction="Fetch data from API 2.")

# This will run fetch_api1 and fetch_api2 at the same time.
gather_concurrently = ParallelAgent(
    name="ConcurrentFetch",
    sub_agents=[fetch_api1, fetch_api2]
)

While there isn't a direct parameter to control the concurrency level within a single ParallelAgent, if you need to process a large number of tasks while respecting rate limits, you could implement a pattern in your application logic to batch the agents. For example, you could group your sub-agents into smaller lists and run them through ParallelAgent sequentially.

I hope this helps clarify the functionality of ParallelAgent!

---

References:
[1] https://google.github.io/adk-docs/agents/multi-agents/
[2] https://google.github.io/adk-docs/agents/workflow-agents/parallel-agents/