How to Automate Computer Tasks With OpenCode

TL;DR

OpenCode can automate browser research, file management, data analysis, visualization, and custom research directly from a terminal. Install it from opencode.ai, authenticate with a supported AI provider, and optionally connect Chrome DevTools through an MCP server. Because it can modify local files and operate a browser, review targets carefully before authorizing destructive actions.

Transcript

Open Code is an open source AI agent that lives in your terminal and can do anything on your machine and it works with any AI model and you can give it any task. In this video, I will show you eight insane use cases that save me many hours every week. From finding leads to organizing files to data analysis to deep research, Open Code can do all of ... Read More

Key Insights

• OpenCode is an open-source AI agent that runs in a terminal, works with multiple AI model providers, and can perform tasks on the user’s machine. After installation and authentication, it can be launched from a global terminal or from an IDE terminal.
• OpenCode installation begins with the terminal command provided at opencode.ai. Running that command installs the agent or updates an existing installation, while the login command displays available AI providers and starts the account authorization process.
• Chrome DevTools MCP gives OpenCode browser-operating capabilities. The demonstrated setup adds the required MCP configuration to the OpenCode JSON file, after which a newly launched terminal session reports that Chrome DevTools is connected.
• Lead generation works by asking OpenCode to research suitable businesses, collect public contact information, and save the findings in a clear file. The demonstration produced 33 Warsaw business leads, including addresses and websites, after searching multilingual web pages.
• Safe file cleanup starts with inspection instead of immediate deletion. OpenCode was first instructed to identify the 20 oldest and 20 largest files in the Downloads folder without deleting anything, allowing the user to review the proposed targets before authorizing removal.
• OpenCode can perform narrowly scoped file deletion when given explicit boundaries. After being told to delete only the identified oldest and largest files, it reduced the demonstrated Downloads folder from 20 GB to 4.3 GB and reported freeing about 15 GB.
• Data visualization can combine analysis, categorization, programming, and image generation in one workflow. OpenCode analyzed 234 startup ideas, organized them into 21 categories, wrote a Python script, installed Matplotlib in a virtual environment, and produced an Obsidian-style PNG graph.
• Complex tasks become manageable when OpenCode creates a to-do list and completes smaller steps sequentially. The demonstrations show it researching, updating task status, writing files, installing dependencies, verifying outputs, and then presenting a concise final result.

Questions & Answers

Q: How do you install and start OpenCode?

Go to opencode.ai, copy the provided installation command, paste it into a terminal, and press Enter. The command installs OpenCode or updates it if it is already present. Next, run the OpenCode login command, select an available AI provider, and complete authorization through the generated link. After authentication succeeds, type open code in any terminal to launch the agent.

Q: How do you connect OpenCode to a browser?

Add the Chrome DevTools MCP configuration to the OpenCode JSON file located in the OpenCode configuration directory. The demonstration asks the agent to create the file if it does not exist or edit it without changing anything else. After saving the configuration, open a new terminal session and launch OpenCode. A successful setup displays that the Chrome DevTools MCP server is connected.

Q: How can OpenCode find leads for a local business?

Give OpenCode a short description of the business, its location, the types of customers that might benefit, and the information you want saved. The demonstrated prompt requested suitable Warsaw businesses for AI and n8n automation, along with online contact details in a Markdown file. Using browser control, OpenCode searched local businesses and produced 33 leads with addresses and websites in 101 seconds.

Q: How can OpenCode safely organize a Downloads folder?

Begin with a read-only inspection request. Ask OpenCode to locate the Downloads folder, identify the oldest and largest files, and present them clearly without deleting anything. Review the results before issuing a second, tightly scoped deletion instruction. In the demonstration, the agent listed the 20 oldest and 20 largest files first, then deleted only those approved targets.

Q: How much storage did OpenCode free in the file cleanup example?

The demonstrated Downloads folder initially contained about 20 GB across 2,400 items. After OpenCode identified and then deleted the approved sets of old and large files, it reported freeing around 15 GB. A folder information check showed that the remaining size was 4.3 GB, confirming a substantial reduction while the deletion instruction was limited to the selected files.

Q: How can OpenCode turn a large idea list into a visual graph?

Ask OpenCode to read the source file, analyze every entry, group related ideas into categories, and export the result as a PNG graph. In the demonstration, it wrote roughly 400 lines of Python, installed Matplotlib, created a virtual environment, generated the visualization, and verified the output. It categorized 234 startup ideas into 21 groups in an Obsidian-style graph.

Q: Why does OpenCode create to-do lists for complex tasks?

OpenCode uses to-do lists to divide a broad request into smaller actions and focus on one step at a time. During the demonstrations, those steps included researching information, categorizing content, writing code, installing dependencies, producing an output file, and verifying that the file existed. This structure makes progress visible and helps the agent complete multi-part requests in an organized sequence.

Q: What safety precautions should you use with OpenCode?

Treat OpenCode as a powerful agent because it can operate a browser, inspect local data, create files, install dependencies, and delete selected files. Use it ethically, request inspection before modification, review proposed targets, and define exact limits for destructive actions. The cleanup example first prohibited deletion, then authorized removal of only the 20 oldest and 20 largest files and nothing else.

Summary & Key Takeaways

• OpenCode is presented as an open-source terminal agent that can work with different AI models and perform tasks on a computer. The setup involves running the installation command from opencode.ai, authenticating through a supported provider, and launching the agent by typing open code in a global terminal or an IDE terminal.

• Browser control is enabled by adding the Chrome DevTools MCP configuration to the OpenCode JSON file. With that connection active, the agent searches websites, understands multilingual pages, gathers business information, and saves results locally. In the demonstration, it collected 33 Warsaw leads with addresses and websites in 101 seconds.

• OpenCode also inspects and changes local files, so cautious prompting is essential. It identified large and old downloads before receiving separate permission to delete selected items, reducing the folder from 20 GB to 4.3 GB. It also categorized startup ideas and generated an Obsidian-style PNG graph through a Python script.