Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Evincere/klisk/llms.txt

Use this file to discover all available pages before exploring further.

Agents are the core building blocks in Klisk. An agent is an AI system that can reason, use tools, and interact with users to accomplish tasks.

​
Defining an agent

Use define_agent() to create and register an agent: 
from klisk import define_agent, get_tools

agent = define_agent(
    name="Assistant",
    instructions="You are a helpful assistant.",
    model="gpt-5.2",
    tools=get_tools("greet", "search"),
)
define_agent() is a thin wrapper around the OpenAI Agents SDK Agent class that automatically registers your agent in the global AgentRegistry so the Studio and CLI can discover it.

​
Parameters

​
name

​
name
string
required
The agent’s unique identifier. Used in the Studio UI and for referencing the agent programmatically.
agent = define_agent(
    name="ResearchAssistant",
    instructions="...",
)

​
instructions

​
instructions
string
The system prompt that defines the agent’s behavior, personality, and capabilities. This guides how the agent interprets user requests and uses tools.
agent = define_agent(
    name="CodeReviewer",
    instructions="""You are an expert code reviewer.
    Analyze code for bugs, security issues, and best practices.
    Provide constructive feedback with specific examples.""",
)
Use multi-line strings for complex instructions. Be specific about the agent’s role, constraints, and output format.

​
model

​
model
string
default:"gpt-5.2"
The language model to use. Supports OpenAI models and any LiteLLM-compatible provider.
See the Models page for detailed information about model configuration and multi-provider support.

​
tools

​
tools
list
A list of tools the agent can use. Pass FunctionTool objects or use get_tools() to retrieve registered tools by name.
from klisk import define_agent, get_tools

agent = define_agent(
    name="Assistant",
    tools=get_tools("greet", "search", "analyze"),
)
See the Tools page for detailed information about creating and using tools.

​
builtin_tools

​
builtin_tools
list
Enable provider-hosted tools like web search and code interpreter. Only supported with OpenAI models.
agent = define_agent(
    name="Researcher",
    model="gpt-5.2",
    builtin_tools=["web_search", "code_interpreter"],
)
See the Builtin Tools page for available tools and configuration options.

​
temperature

​
temperature
float
Controls randomness in responses. Range: 0.0 (deterministic) to 2.0 (very random). Defaults to the model’s default.
agent = define_agent(
    name="CreativeWriter",
    temperature=1.5,  # More creative
)

agent = define_agent(
    name="DataAnalyst",
    temperature=0.2,  # More focused
)

​
reasoning_effort

​
reasoning_effort
string
Controls how much reasoning the model applies. Only for o-series models (o1, o3, o4-mini) and gpt-5+ (gpt-5.1, gpt-5.2).
Valid values: "none", "minimal", "low", "medium", "high", "xhigh" 
agent = define_agent(
    name="Analyst",
    model="o3",
    reasoning_effort="high",
)
LiteLLM automatically translates reasoning_effort to each provider’s equivalent parameter.
Only set reasoning_effort for reasoning models. It has no effect on standard models like gpt-4o or claude-sonnet-4-5.

​
Agent lifecycle

When you call define_agent():
  1. Model resolution — The model string is parsed and resolved to either a native OpenAI model or a LitellmModel instance
  2. Tool registration — All specified tools are validated and attached to the agent
  3. SDK agent creation — An underlying Agent instance from the OpenAI Agents SDK is created
  4. Registry registration — The agent is registered in the global AgentRegistry with metadata about its configuration
The registry enables:
  • Studio discovery — The Klisk Studio UI automatically detects and displays your agent
  • CLI access — Commands like klisk run and klisk check can reference your agent
  • Hot reload — Changes to your code update the Studio in real-time

​
Source file tracking

define_agent() automatically captures the source file where it’s called using inspect.stack(). This metadata is used by:
  • Studio — Shows which file defines each agent
  • Hot reload — Watches the correct files for changes
  • Error messages — Provides accurate file locations in diagnostics
# src/main.py
from klisk import define_agent

agent = define_agent(name="Assistant")  # Tracked as src/main.py:4

​
Complete example

from klisk import define_agent, get_tools

# Main assistant with custom tools
assistant = define_agent(
    name="Assistant",
    instructions="""You are a helpful assistant that can search the web,
    analyze data, and help with coding tasks.""",
    model="gpt-5.2",
    temperature=0.7,
    tools=get_tools("search", "analyze", "format_code"),
    builtin_tools=["web_search"],
)

# Specialized reasoning agent
analyst = define_agent(
    name="DataAnalyst",
    instructions="Analyze data and provide insights.",
    model="o3",
    reasoning_effort="high",
    temperature=0.3,
    tools=get_tools("load_csv", "plot_graph"),
)

# Multi-provider agent using Anthropic
researcher = define_agent(
    name="Researcher",
    instructions="Research topics and summarize findings.",
    model="anthropic/claude-sonnet-4-5-20250929",
    tools=get_tools("search", "summarize"),
)

​
Next steps

Tools

Learn how to create custom tools for your agents

Models

Explore multi-provider support and model configuration

Build docs developers (and LLMs) love

Get started for freeTalk to us