# AI use cases for Openbot (nightly-labs/openbot) Repo: [nightly-labs/openbot](https://github.com/nightly-labs/openbot) | Use cases: 4 | Page: https://dedupedai.com/use-cases/nightly-labs-openbot/ OpenBot is a local-first desktop app and optional always-on server for running persistent AI coding and assistant agents, each with its own workspace, queue, and conversation. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Run persistent agents on a headless server: https://dedupedai.com/use-cases/nightly-labs-openbot/run-always-on-agent-server/ - Split a coding task across Codex and Claude agents: https://dedupedai.com/use-cases/nightly-labs-openbot/split-coding-work-across-agents/ - Have an agent research and test sites in its own browser: https://dedupedai.com/use-cases/nightly-labs-openbot/delegate-browser-research-to-agent/ - Run an agent on a local model with Ollama or LM Studio: https://dedupedai.com/use-cases/nightly-labs-openbot/run-agent-on-local-model-server/ ## How to run persistent agents on a headless server with Openbot Repo: [nightly-labs/openbot](https://github.com/nightly-labs/openbot) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Install OpenBot on a VPS or home server so agents keep running when your laptop is off. **Who it is for:** A developer or small team lead with a spare Linux VPS or home server who wants agents available at all times, not tied to one desktop. **The problem:** Desktop agents stop when the laptop sleeps or closes, so queued work and long threads stall. **The outcome:** An OpenBot server runs under your account and you reach the same agents from the desktop app, the iPhone app, or Openbot.run/app. ### Steps 1. Use an Ubuntu 24.04 VPS or home server that runs systemd, and open a terminal on it. 2. Run `curl -fsSL https://raw.githubusercontent.com/nightly-labs/openbot/main/scripts/install-server.sh | sudo bash` 3. Run `sudo openbot login` and enter the one-time email code. 4. Open the desktop app, the iPhone app, or Openbot.run/app and confirm the agents appear. 5. Read the self-hosted servers guide for the available options and `openbot` commands before changing defaults. ### Prompt to paste into your agent ```text In the nightly-labs/openbot repo, install the always-on server on following docs/self-hosted-server.md. Expected output: the result of `sudo openbot login` showing a completed sign-in and a list of agents reachable from the desktop app. Verify the sign-in finished before reporting success. ``` ### Caveat Server agents run with danger-full-access and approvalPolicy never, so only host them on machines you are willing to let agents control. Risk, remote script: The install command pipes a script from GitHub straight into sudo bash, so review the script first if you do not trust the source. 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/nightly-labs-openbot/run-always-on-agent-server/ ## How to split a coding task across Codex and Claude agents with Openbot Repo: [nightly-labs/openbot](https://github.com/nightly-labs/openbot) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Give each agent its own workspace and queue, then delegate parts of a task between them. **Who it is for:** A software engineer with a multi-part feature who wants Codex and Claude Code to work in parallel and hand results to each other. **The problem:** One agent session loses context when a large task spans several areas, and manual copy-paste between sessions is slow. **The outcome:** Two persistent agents each hold a dedicated workspace and conversation, and they exchange messages and files inside one desktop app. ### Steps 1. Install the Codex CLI with `curl -fsSL https://chatgpt.com/codex/install.sh | sh` and the Claude CLI with `curl -fsSL https://claude.ai/install.sh | bash`, or use the Download option in OpenBot. 2. Download the OpenBot release for your system from GitHub Releases and open it. 3. Create one Codex agent and one Claude agent, each with its own workspace. 4. Queue a task for each agent, then send an agent-to-agent message asking one to review the other's output. 5. Open a shared channel chat, assign one task owner, and use Stop or Reassign if a delegation goes wrong. ### Prompt to paste into your agent ```text In the nightly-labs/openbot repo, set up a Codex agent and a Claude agent in the desktop app, each with a separate workspace under . Expected output: both agents listed with queued tasks and one agent-to-agent message sent. Verify that each agent's queue shows the task before reporting success. ``` ### Caveat Claude subscriptions fall under provider terms, which OpenBot links but does not change, so check those terms before running heavy parallel work. 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/nightly-labs-openbot/split-coding-work-across-agents/ ## How to have an agent research and test sites in its own browser with Openbot Repo: [nightly-labs/openbot](https://github.com/nightly-labs/openbot) | Difficulty: Beginner | Time: 30 minutes | Basis: From the README Let a persistent agent open, inspect, and control the embedded browser for web research or QA. **Who it is for:** A product or QA person who needs repeated checks of a web flow or a list of pages and wants an agent to do the clicking and reading. **The problem:** Manual page checks are repetitive, and screenshots scattered across tools are hard to compare. **The outcome:** An agent opens the pages in OpenBot's persistent embedded browser, inspects them, and reports what it found in its conversation. ### Steps 1. Install and open the OpenBot desktop app, then complete the first-launch consent. 2. Create an agent and select a model and profile in its settings. 3. Ask the agent in chat to open the target URLs in the embedded browser and inspect each page. 4. Review the agent's report and any files it shared through managed file transfers. 5. Archive the thread when the check is complete, or keep it to rerun later. ### Prompt to paste into your agent ```text In the nightly-labs/openbot repo, create an agent in the desktop app and ask it to open each URL in in the embedded browser and report the page title and any visible errors. Expected output: a per-URL list of results in the agent's conversation. Verify that every URL was opened before reporting success. ``` ### Caveat The agent controls the browser without per-action confirmations, so avoid pointing it at accounts or pages you cannot afford to change. Risk, credentials: If the agent logs into a site in the embedded browser, any credentials it sees stay in that browser session, so use test accounts only. 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/nightly-labs-openbot/delegate-browser-research-to-agent/ ## How to run an agent on a local model with Ollama or LM Studio with Openbot Repo: [nightly-labs/openbot](https://github.com/nightly-labs/openbot) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Connect an agent to a local model server so prompts stay on your machine. **Who it is for:** A developer who wants to try agents without sending prompts to a hosted provider, or who already runs Ollama or LM Studio. **The problem:** Hosted providers are not an option for some code or data, and the agent setup needs to work offline from the cloud. **The outcome:** An agent backed by a local model server, with the model chosen in OpenBot's model picker. ### Steps 1. Start Ollama or LM Studio on the same computer and load a model. 2. Open OpenBot and go to agent setup or the provider settings. 3. Let OpenBot detect the local model server, or add a custom OpenAI-compatible endpoint if detection does not find it. 4. Create an agent that uses the detected model and send it a test prompt. 5. Check the agent's reply in its conversation before giving it real work. ### Prompt to paste into your agent ```text In the nightly-labs/openbot repo, connect the desktop app to the local model server at and create an agent that uses it. Expected output: one test reply from the agent in its conversation. Verify the reply came from the local server before reporting success. ``` ### Caveat Local models may not match hosted models for tool use or long threads, and OpenBot still runs the agent with danger-full-access. 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/nightly-labs-openbot/run-agent-on-local-model-server/