OpenCode, a new open‑source AI coding platform, has impressed developers with its flexible agent and workflow system. Built on a TUI console, it supports over 75 LLM providers, including BedRock, Gemini, and a user’s own Claude Pro subscription, letting teams switch models without vendor lock‑in. Unlike Claude Code, which relies on a single CLI and Markdown‑only configuration, OpenCode offers JSON/Markdown settings and JSON‑based manifests that flag errors before they cascade.

The platform ships a library of built‑in tools and allows custom MCP servers—local or remote—via a simple `opencode.jsonc` file. Commands automate repetitive tasks such as running tests, generating PRDs, or updating README files, while agents specialize in file organization, project analysis, diagram generation, copywriting, and technical documentation. An orchestrator agent can chain these sub‑agents into a full documentation pipeline, logging progress and enabling recovery from failures.

The open‑source repo on GitHub demonstrates a complete workflow that can be adapted to any project. As AI tooling matures, OpenCode’s agnostic, modular approach positions it as a versatile alternative for teams seeking greater control and cost efficiency.