Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.sudoiq.com/llms.txt

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

The Agents API helps you run and optimize multi-step agents—versioned workflows that call models, tools, and validators to complete a task end-to-end.

​
sudoiq at a high level

If you are new here, think of it as four steps which would take about 15-20 min for a new agent:
  1. Define an agent in sudoiq with your account. That definition is the full recipe: which tools the model may call, what you want the workflow to accomplish, and how the steps connect. Use the WYSIWYG-style visual editor in the product, or chat with an authoring agent that can draft and revise the graph for you. You also pick at least one model provider (OpenAI, Anthropic, etc.) and input an API key for sudoiq to use when calling your agent. The agent it builds for you is yours and fully portable, you can move it off sudoiq later and self-host it if needed.
  2. Embed the client library in your application (today that is mainly the Python SDK) or script against the HTTPS API with the agentservice CLI. This layer starts runs, returns tool results when the model pauses, registers schemas if you manage them from code, and sends feedback when you want humans to grade a run.
  3. Run the agent with real inputs from your app or a script. sudoiq runs the workflow on its side: model calls, tool traces, validators, and any graders or eval hooks you turned on.
  4. sudoiq optimization agents can keep improving what you ship. They read run history, grader scores, tests, and feedback, then propose or stage changes to prompts, tools, or graph structure. You can try those ideas safely—for example by replaying old runs or running small A/B comparisons—before you adopt them for everyone.
Sign up at sudoiq, create an API key under Settings, and integrate in minutes with Python or the agentservice CLI.

​
Example: one run, then evals and feedback

A minimal pattern: start a run from your app. The platform attaches task identity; record user feedback on that task so optimization agents can use it with other runs. Exact symbols match your installed agentsapi package; see Getting started. 
import asyncio
from agentsapi import AgentServiceAPIClient

async def main():
    client = AgentServiceAPIClient()
    client.set_api_key("sk_...")  # https://sudoiq.com/settings

    # runs and awaits until done, calls any tools 
    result = await client.run_agent(
        agent_id="YOUR_AGENT_UUID",
        string_inputs={"1": "Summarize the Q3 memo.", "2": "/file/url.pdf"},
        environment="production",
    )
    file.write(result)

asyncio.run(main())
This example is extremely basic. You can build multi-turn agents that use tools, have intelligent retries, etc. The code is still simple for executing the agent. Most of the tests, evaluations, feedback, and validation handling are managed within sudoiq and not kept in code. They are accessible through the CLI, API, or UI.

​
Libraries and tools

The Python client (import agentsapi) for running agents. The Command Line Interface (agentservice) handles agent editing, execute, tasks, tools and validators push, and demos. Easy to integrate into Claude Code or Cursor. The Node.js SDK is coming soon; see Node.js client. Self-hosting is optional for sensitive data or dedicated environments: some customers deploy the API stack on their own infrastructure with Docker and managed Postgres and Redis (Self-hosted Docker).

​
Embedding in your app (Python / CLI)

Python 3.10+ is recommended for the CLI and examples. Dependencies ship with the client (httpx, pydantic).

​
Next steps