# Agent

Agent is a self-hosted ai app. OpenAI Agents SDK is a lightweight, provider-agnostic Python framework for building multi-agent AI workflows.

Canonical page: https://ourlocalhost.com/app/5069/agent

## Description

OpenAI Agents SDK is a lightweight, provider-agnostic Python framework for building multi-agent AI workflows. It defines agents with instructions, tools, guardrails, and handoffs, and works with the OpenAI Responses and Chat Completions APIs plus 100+ other large language models, letting developers compose and orchestrate reliable agentic applications.

## Platform availability

Available on 1 of 10 platforms: Helm.

## Health score

72/100 (Good), based on 5 of 5 factors.
- Maintained: 100 (last commit 15 days ago)
- Popular: 90 (30,532 GitHub stars)
- Easy to install: 35 (1 install platform, install notes, Docker image)
- Light to run: 100 (128 MB minimum RAM)
- Security check: 10 (3 serious findings, 6 minor, CI, tests)

## Security check

3 serious findings: Uses litellm 1.83.0, which has a known critical vulnerability; Uses litellm 1.83.0, which has a known critical vulnerability; Uses litellm 1.83.0, which has a known critical vulnerability.
Automated checks on commit 125efa0, 12 Oct 2026. Not an audit.

## Links

- GitHub: https://github.com/openai/openai-agents-python
- Official website: https://openai.github.io/openai-agents-python/

## Common questions

### Do I need to run a setup wizard before starting an agent?

Yes, run docker agent setup before starting an agent and choose a model provider and model. The agent needs to know where AI models come from to answer requests.

### What port must a local model server use for the agent to reach it?

Start the local model server on port 8000, such as with vLLM. If port 8000 is not running or is blocked, the agent cannot reach that model.

### How can I verify a harness integration's credentials before running it?

Run docker agent doctor ./agent.yaml and then docker agent setup to check installation and login status. This catches missing credentials or setup steps before the agent tries to run.

### What kinds of problems are common in self-hosted agent systems?

Common problems involve GPU memory, model loading, inference performance, container management, and agent behavior. Most issues have straightforward solutions once the root cause is identified.

---
Source: Our Local Host (https://ourlocalhost.com/app/5069/agent), the self-hosted and open-source app directory.
