> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agno.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Dynamic Agents
> Build agents, teams, and workflows per request from JWT claims, user input, and other request-time context.
v2.6.0
Build a fresh Agent, Team, or Workflow for each incoming request. A factory is a callable that AgentOS invokes per request, so tools, instructions, model, and database scope can depend on who's calling.
```python tenant_factory.py theme={null}
from agno.agent import Agent, AgentFactory
from agno.db.postgres import PostgresDb
from agno.factory import RequestContext
from agno.models.openai import OpenAIResponses
from agno.os import AgentOS
db = PostgresDb(
id="factory-demo-db",
db_url="postgresql+psycopg://ai:ai@localhost:5532/ai",
)
def build_tenant_agent(ctx: RequestContext) -> Agent:
user_id = ctx.user_id or "anonymous"
return Agent(
model=OpenAIResponses(id="gpt-5.4"),
db=db,
instructions=f"You are a helpful assistant for tenant {user_id}. Be concise.",
markdown=True,
)
tenant_factory = AgentFactory(
id="tenant-agent",
db=db,
factory=build_tenant_agent,
name="Per-tenant assistant",
description="Builds a personalized agent per tenant on each request.",
)
agent_os = AgentOS(agents=[tenant_factory])
app = agent_os.get_app()
if __name__ == "__main__":
agent_os.serve(app="tenant_factory:app", port=7777)
```
Hit `POST /agents/tenant-agent/runs` and the factory is invoked with a [`RequestContext`](/reference/agent-os/factories#requestcontext) for that request.
`id` and `db` are required on `AgentFactory`. The factory's `id` overrides any `id` set on the `Agent` returned by the callable.
## When to Use a Factory
Use a plain `Agent` / `Team` / `Workflow` when the component is shared across all callers. Reach for a factory when construction depends on the request.
| Use Case | Pattern |
| ------------------------------------- | ------------------------------------------------------------ |
| Tools or model vary per caller role | `AgentFactory` reading `ctx.trusted.claims` |
| Members vary per tenant | `TeamFactory` returning a `Team` with tenant-specific agents |
| Pipeline shape varies per request | `WorkflowFactory` returning a different step graph |
| Client picks persona, depth, or style | Factory with `input_schema` reading `ctx.input` |
`GET /agents/{id}`, `GET /teams/{id}`, and `GET /workflows/{id}` (the component-detail endpoints) return the factory's metadata without invoking it. See the [Factories reference](/reference/agent-os/factories) for per-endpoint behavior and discovery payload shape.
## Learn How To
Produce a fresh Agent per request from verified claims or client input.
Compose a fresh Team and its members for each request.
Compose pipelines whose step graph depends on the request.
Fields, sources, and identity precedence.
Populate `ctx.trusted.claims` from verified JWTs.
Class signatures, parameters, and exceptions.
## Developer Resources
* [Factory examples](/examples/agent-os/factories/overview)
* [AgentFactory / TeamFactory / WorkflowFactory reference](/reference/agent-os/factories)
* [RequestContext reference](/reference/agent-os/factories#requestcontext)