Simile

raw JSON →

0.5.4 verified Fri May 01 auth: no python

A Python SDK for interfacing with Simile AI agents in simulations. Version 0.5.4, pre-1.0 rapid development.

pip install simile

Common errors

error ImportError: cannot import name 'Agent' from 'simile.agent' ↓

cause Import path changed in v0.5.0.

fix

Use from simile import Agent.

error AuthError: No API key provided ↓

cause Missing or empty SIMILE_API_KEY environment variable.

fix

Set the environment variable SIMILE_API_KEY or pass api_key to Agent constructor.

error TypeError: Agent.__init__() got an unexpected keyword argument 'model' ↓

cause `model` parameter removed in 0.6.0 (future).

fix

Use AgentConfig(model='gpt-4') and pass as config to Agent.

Warnings

breaking In v0.5.0, the import paths were flattened. `from simile import Agent` replaces `from simile.agent import Agent`. Old imports will break. ↓

fix Use top-level imports: `from simile import Agent, Simulation`.

gotcha API key must be set; either via `SIMILE_API_KEY` environment variable or `api_key` parameter. If missing, SDK will raise an `AuthError`. ↓

fix Set `SIMILE_API_KEY` env var or pass `api_key` explicitly.

deprecated The `model` parameter in `Agent` is deprecated since 0.5.2. Use `agent_config` with `AgentConfig` instead. ↓

fix Use `AgentConfig(model='gpt-4')` and pass it to `Agent(config=...)`.

Imports

Agent
wrong
```
from simile.agent import Agent
```
correct
```
from simile import Agent
```
Top-level import changed in 0.4.0; submodule import no longer works.

Simulation

wrong

from simile.simulation import Simulation

correct

from simile import Simulation

Simulation is now exported at package level.

AgentConfig
```
from simile import AgentConfig
```
Used to configure agent personality and knowledge base.

Quickstart

Initialize a Simile agent and run a simple simulation.

from simile import Agent, Simulation

# Create an agent
agent = Agent(
    name="CustomerBot",
    api_key=os.environ.get("SIMILE_API_KEY", ""),
    model="gpt-4"
)

# Create a simulation
sim = Simulation(agents=[agent])

# Run a conversation
response = sim.run("Hello, how are you?")
print(response)