When to Use LiteAgent
- ✓Embedded agents inside larger applications.
- ✓Quick utilities exposed via an API endpoint.
- ✓Sub-steps inside Flows that don't need delegation.
- ✗Multi-agent collaboration — use a Crew with
allow_delegation=True.
Basic Use
from crewai import LiteAgent
agent = LiteAgent(
role="Summarizer",
goal="Produce a 3-bullet summary of any input text",
backstory="You distill complex prose into the essential points.",
tools=[],
)
out = agent.run("Long article text here ...")API: LiteAgent.
Inside a Flow
LiteAgents shine as sub-steps in larger Flows. Parent-flow identification flows through automatically so traces nest correctly.
@listen(fetch_data)
def summarize(self, raw):
return LiteAgent(role="Summarizer", goal="...", backstory="...").run(raw)Events
The runtime emits LiteAgentExecutionStartedEvent, LiteAgentExecutionCompletedEvent, and LiteAgentExecutionErrorEvent — same lifecycle plumbing as a full Crew.
Notes
LiteAgent is not a license to skip contracts
Smaller entry points still need clear instructions and output expectations. Without them, you simply moved brittle prompting from a crew definition into ad hoc strings.
Compose inside Flow for mixed workloads
Flows handle orchestration while LiteAgent handles single-shot reasoning. Keep Flow state explicit so retries do not duplicate side effects when a LiteAgent step partially succeeds.
Utilities should stay pure when possible
Helper functions that mutate global config or secrets make tests flaky. Pass dependencies explicitly so LiteAgent calls remain parallel-safe in workers.
Know when to graduate back to a crew
When you add memory, multiple roles, and branching reviews, the ergonomics of a full crew usually beat stitching many LiteAgent calls by hand.
CrewAI LiteAgent FAQ
What is CrewAI LiteAgent?
LiteAgent is a streamlined path for single-agent LLM work with CrewAI ergonomics when you do not need multi-agent scheduling or a full crew graph.
When should I pick LiteAgent over a Crew?
Pick LiteAgent for one-shot classification, summarization, or tool calls inside a larger system, and upgrade to a crew when roles and handoffs become explicit.
Can LiteAgent live inside a CrewAI Flow?
Yes. Flow methods often call LiteAgent for small steps while delegating heavier multi-agent work to embedded crews.
Does LiteAgent support tools?
Yes, when configured similarly to other CrewAI agents. Keep toolsets small to preserve the simplicity that motivated LiteAgent in the first place.
How does LiteAgent affect costs?
It can lower cost versus spinning a full crew when one strong prompt plus tools suffices, but you still pay for tokens and tool usage on each call.
See also: DevShelfHub's CrewAI tool review for a product-level comparison, pricing notes, and links back into this tutorial series.