docs: add AgentField to Agent SDKs integrations#22901
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Greptile SummaryThis is a documentation-only PR that adds AgentField — an open-source control plane for autonomous AI agents — to LiteLLM's integrations page and Agent SDKs sidebar. It introduces a new tutorial (
Confidence Score: 4/5
|
| Filename | Overview |
|---|---|
| docs/my-website/docs/tutorials/agentfield.md | New tutorial for AgentField integration; contains a top-level await in the Quick Start snippet (SyntaxError in script context) and two unused JSX imports (Tabs, TabItem). |
| docs/my-website/docs/integrations/index.md | Adds AgentField entry to the AI Agent Frameworks section and fixes missing newline at end of file. No issues found. |
| docs/my-website/sidebars.js | Adds tutorials/agentfield as the first entry in the Agent SDKs sidebar category. No issues found. |
Sequence Diagram
sequenceDiagram
participant User
participant AgentField SDK
participant LiteLLM
participant LLM Provider
User->>AgentField SDK: agent.run("prompt")
AgentField SDK->>LiteLLM: litellm.acompletion(model, messages)
LiteLLM->>LLM Provider: API request (OpenAI / Anthropic / Azure / Ollama / …)
LLM Provider-->>LiteLLM: LLM response
LiteLLM-->>AgentField SDK: Normalised response
AgentField SDK-->>User: Agent response
Last reviewed commit: 2881024
| ```python | ||
| from agentfield import Agent, AgentConfig | ||
|
|
||
| config = AgentConfig( | ||
| name="my-agent", | ||
| model="gpt-4o", # Any LiteLLM-supported model | ||
| instructions="You are a helpful assistant." | ||
| ) | ||
|
|
||
| agent = Agent(config) | ||
| response = await agent.run("Hello, world!") | ||
| ``` |
There was a problem hiding this comment.
await used outside an async context
The snippet uses await agent.run(...) at the top level of a code block without wrapping it in an async def function. In Python, await expressions are only valid inside async def functions (or in an async REPL like IPython). Running this code as a plain .py script will raise a SyntaxError.
The example should either wrap the call in an async def main() function and use asyncio.run(), or show the synchronous variant if one exists.
| ```python | |
| from agentfield import Agent, AgentConfig | |
| config = AgentConfig( | |
| name="my-agent", | |
| model="gpt-4o", # Any LiteLLM-supported model | |
| instructions="You are a helpful assistant." | |
| ) | |
| agent = Agent(config) | |
| response = await agent.run("Hello, world!") | |
| ``` | |
| import asyncio | |
| from agentfield import Agent, AgentConfig | |
| config = AgentConfig( | |
| name="my-agent", | |
| model="gpt-4o", # Any LiteLLM-supported model | |
| instructions="You are a helpful assistant." | |
| ) | |
| async def main(): | |
| agent = Agent(config) | |
| response = await agent.run("Hello, world!") | |
| asyncio.run(main()) |
| import Tabs from '@theme/Tabs'; | ||
| import TabItem from '@theme/TabItem'; |
There was a problem hiding this comment.
Unused JSX imports
Tabs and TabItem are imported at the top of the file but are never used anywhere in the document. In Docusaurus this can produce build warnings. These two lines should be removed.
| import Tabs from '@theme/Tabs'; | |
| import TabItem from '@theme/TabItem'; |
d92bb15
into
BerriAI:litellm_oss_staging_03_06_2026
Summary
Adds AgentField to the Agent SDKs section in the integrations docs.
AgentField is an open-source control plane for building and orchestrating autonomous AI agents. Its Python SDK uses LiteLLM internally (
litellm.acompletion()) for multi-provider LLM support, giving agents access to 100+ providers out of the box.Changes
integrations/index.mdunder AI Agent Frameworkstutorials/agentfield.mdwith usage examples (OpenAI, Anthropic, Ollama, Azure, LiteLLM Proxy)tutorials/agentfieldto the Agent SDKs sidebar sectionLinks
cc @AbirAbbas