Descriptions:
Open Code is an open-source terminal AI agent that works with any model and has full access to the host machine. David Ondrej demonstrates eight practical use cases, walking through live prompts and results across lead generation, competitor research, deep content analysis, system diagnostics, and article writing.
The lead generation demo connects Open Code to a Chrome DevTools MCP server, allowing the agent to browse and scrape business contacts in Warsaw, Poland autonomously from a single plain-English prompt. The deep research use case is particularly detailed: given a prompt about reducing appointment no-show rates, Open Code analyzed 13 YouTube videos totaling 75,000 combined views and produced a nearly 300-line report including video-level breakdowns, combined view counts, and an executive summary — completed in five minutes and fifteen seconds. The system diagnostics demo surfaces load averages, memory pressure, swap activity, and top resource consumers (Brave browser, Docker, OBS) without making any changes to the system.
Setup starts at opencode.ai with a single terminal install command. Authentication supports existing ChatGPT Pro, Anthropic, and GitHub subscriptions. The Chrome DevTools MCP server is added by editing opencode.json in the root config directory. Open Code runs inside both a global terminal and IDE-embedded terminals such as Cursor, with model selection (including Opus 4.5 and GPT models) handled at launch.
📺 Source: David Ondrej · Published January 22, 2026
🏷️ Format: Hands On Build
