# AI use cases for Open Agent Core (MiniMax-AI/OpenAgentCore) Repo: [MiniMax-AI/OpenAgentCore](https://github.com/MiniMax-AI/OpenAgentCore) | Use cases: 3 | Page: https://dedupedai.com/use-cases/minimax-ai-openagentcore/ OpenAgentCore is a self-hosted, open source Agents API that runs Codex, Claude Code or MiniMax Code inside sandboxes, using the official OpenAI SDK. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Run a coding agent Session through the OpenAI SDK: https://dedupedai.com/use-cases/minimax-ai-openagentcore/run-coding-agent-sessions-via-openai-sdk/ - Point an existing OpenAI SDK app at self-hosted Core: https://dedupedai.com/use-cases/minimax-ai-openagentcore/point-existing-openai-app-at-self-hosted-core/ - Run agent Sessions on your own Linux, macOS or Windows machine: https://dedupedai.com/use-cases/minimax-ai-openagentcore/run-agent-sessions-on-own-machine/ ## How to run a coding agent Session through the OpenAI SDK with Open Agent Core Repo: [MiniMax-AI/OpenAgentCore](https://github.com/MiniMax-AI/OpenAgentCore) | Difficulty: Intermediate | Time: 60 minutes | Basis: From the README Install Core and Web, then start a Codex Session from Python using the OpenAI SDK. **Who it is for:** A developer who wants to call a coding agent from an application and keep the runtime on their own infrastructure. **The problem:** Hosted agent APIs may not fit where code, models or data must stay under your control. **The outcome:** A self-hosted Core installation that returns a Session ID from an OpenAI SDK call, with the agent able to create files in /workspace. ### Steps 1. Run `curl -fsSL https://openagentcore.dev/install.sh | bash` on Linux or macOS, or `irm https://github.com/MiniMax-AI/OpenAgentCore/releases/latest/download/install.ps1 | iex` in PowerShell on Windows. 2. Open Web, sign in with the Core key the installer prints, set the public address and a default model. 3. In Web, issue a Project API key and add sandbox capacity (a node or E2B). 4. Export the variables: set OPENAI_BASE_URL to your API base URL (for example https://core.example/v1) and OPENAI_API_KEY to the Project API key. (inferred, check the README) 5. Run the Python quickstart from the README, which calls client.beta.agents.sessions.create with harness set to codex, and print session.id. ### Prompt to paste into your agent ```text Using the MiniMax-AI/OpenAgentCore repo, install Core and Web on and configure a default model and Project API key. Then run the README Python quickstart with OPENAI_BASE_URL set to and confirm it prints a Session ID. Verify that the Session finishes and that /workspace/hello.txt exists in the Session output. ``` ### Caveat The README says the install needs Docker and that you must add sandbox capacity (a node or E2B) before Sessions can run. Risk, credentials: The Core key and Project API key grant access to your installation, so store them in a secrets manager and do not commit them to code. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/minimax-ai-openagentcore/run-coding-agent-sessions-via-openai-sdk/ ## How to point an existing OpenAI SDK app at self-hosted Core with Open Agent Core Repo: [MiniMax-AI/OpenAgentCore](https://github.com/MiniMax-AI/OpenAgentCore) | Difficulty: Intermediate | Time: 60 minutes | Basis: From the README Change the base URL and key so an OpenAI SDK app runs agent Sessions on your own Core. **Who it is for:** An engineer with an application already built on the OpenAI SDK who needs to move the backend to infrastructure they control. **The problem:** Rewriting an app for a new API costs time, and the app may need to keep its current SDK calls. **The outcome:** The same SDK code sends agent Sessions to your Core installation, with no change to the request shape. ### Steps 1. Install Core and Web with `curl -fsSL https://openagentcore.dev/install.sh | bash` on Linux or macOS, or the PowerShell command from the README on Windows. 2. In Web, configure the public address, set a default model and issue a Project API key. 3. Set OPENAI_BASE_URL to https://core.example/v1 (your own API base URL) and OPENAI_API_KEY to the Project API key in the app's environment. (inferred, check the README) 4. Add the extra_body agent setting with x_agents_core and harness from the README to choose the harness per Session. 5. Run the app's existing test or a single request and check the returned Session ID. ### Prompt to paste into your agent ```text In the MiniMax-AI/OpenAgentCore repo, follow the README Quickstart to point an existing OpenAI SDK app at a Core installation at . Change only OPENAI_BASE_URL and OPENAI_API_KEY, then run and confirm the request returns a Session ID from the self-hosted endpoint. ``` ### Caveat The README says the Agents API uses the same protocol as OpenAI's Agents API, but it does not list every endpoint, so test the exact calls your app makes. Risk, credentials: Keeping the Project API key in environment variables means anyone with access to that environment can start Sessions on your Core. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/minimax-ai-openagentcore/point-existing-openai-app-at-self-hosted-core/ ## How to run agent Sessions on your own Linux, macOS or Windows machine with Open Agent Core Repo: [MiniMax-AI/OpenAgentCore](https://github.com/MiniMax-AI/OpenAgentCore) | Difficulty: Advanced | Time: 90 minutes | Basis: From the README Add a personal machine as sandbox capacity so Sessions run on hardware you own. **Who it is for:** A developer or operator who wants agent work to run on a machine they control instead of a hosted sandbox. **The problem:** Hosted sandbox capacity may be unavailable or may not be where local tools, files or models live. **The outcome:** Sessions that execute on your own machine, configured through Web as sandbox capacity. ### Steps 1. Read the self-hosted guide at https://openagentcore.dev/docs/getting-started/self-hosted and check the prerequisites for your operating system. 2. Install Core and Web with `curl -fsSL https://openagentcore.dev/install.sh | bash` on Linux or macOS, or the PowerShell command from the README on Windows. 3. In Web, sign in with the Core key, then add your own machine as sandbox capacity. 4. Start a Session with the Python quickstart and set the harness to codex, claude-code or the MiniMax Code harness the README names. 5. Check the Session output and confirm the files were created on your machine. ### Prompt to paste into your agent ```text Using the MiniMax-AI/OpenAgentCore repo, set up Core and Web on , add that machine as sandbox capacity in Web, and start one Session that writes a file to /workspace. Report the Session ID and confirm the file exists on the machine after the Session completes. ``` ### Caveat The README says idle sandboxes resume with files intact only on microsandbox with independent workspace storage, so other providers may not keep workspaces the same way. Risk, remote script: The install step pipes a remote script into bash or iex, so review the script or pin the release before running it on a machine you care about. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/minimax-ai-openagentcore/run-agent-sessions-on-own-machine/