Overview
Every public name in guardian_agent and its framework modules, generated from the SDK's docstrings.
The reference covers every public name of the Guardian Agent SDK, generated from its docstrings, so it matches the current code. Import the package as guardian_agent; the examples write it ga.
An enrolled agent, described and routed through the local Viper proxy, so that every model request it makes is signed as that agent:
python
from openai import OpenAI
import guardian_agent as ga
guardian = ga.Guardian() # the proxy's address, from VIPER_PROXY_URL
researcher = guardian.agent(
"researcher", # as enrolled with `viper agent add researcher`
model=ga.Model("openai", "gpt-4.1", temperature=0),
instructions="You research questions and cite your sources.",
)
client = researcher.bind(OpenAI()) # every request now signed as the researcher
reply = client.chat.completions.create(
model="gpt-4.1",
temperature=0,
messages=[{"role": "user", "content": "What changed in the last release?"}],
)
print(reply.choices[0].message.content)| Page | What it covers |
|---|---|
| Guardian and Agent | Guardian, the connection to the local Viper proxy, and Agent, an enrolled agent you route model clients through |
| Definition | Definition, what an agent is, with SubAgent and Problem |
| Models and prompts | Model, Reasoning and Thinking; the System and Developer prompts |
| Tools | Tool, the @tool decorator, and McpServer |
| Skills and memory | Skill and Memory |
| Framework and code | Framework, Middleware, Graph and Source |
| Capabilities | Capabilities and Limits, what a proxy supports |
| Errors | guardian_agent.errors, every error the SDK raises, and proxy_error |
| LangChain | guardian_agent.langchain: create_agent and agent_of |
| LangGraph | guardian_agent.langgraph: describe_graph |
| Deep Agents | guardian_agent.deepagents: create_deep_agent |
| Testing | guardian_agent.testing: FakeProxy and FakeRequest |