Anthropic takes legal action against OpenCode

Backlash erupts as Anthropic forces OpenCode to drop Claude — “Why fight open source?”

TLDR: OpenCode stripped Claude features after legal pressure from Anthropic, removing logins, prompts, and a beta flag. The dev community erupted, accusing Anthropic of hostility to open source, warning of trust issues, and drawing Google comparisons, while a few argued it’s just a company enforcing its rules.

Anthropic told OpenCode to scrub Claude from the app — the Claude login option is gone, the plugin is yanked, the branded prompt file is deleted, even a special “beta” header is dropped. Docs now say no Anthropic OAuth logins, and one step still mentions an option that no longer exists. Bonus drama: a refactor may have accidentally removed a “User-Agent” tag for other AI providers, which devs flagged as a possible regression.

The internet? Absolutely on fire. The top vibe is “why pick a fight with open source?” with one user sighing that Anthropic is “on a roll making friends.” Another jabbed, “It’s ok when Anthropic do it,” hinting at bigger ethical gripes. A cynical hot take claimed Anthropic prefers lawyers to partnerships — “suing, not buying.” And in the background, folks dug up a previous thread about Anthropic blocking OpenCode, fueling the narrative that this wasn’t a one-off.

Comparisons to Google’s past account bans spooked people: “Hopefully it doesn’t come to that with Anthropic.” Memes flew hard: “villain arc speedrun,” “Pro/Max → Pro/Nope,” and the pun of the day, lawyers > layers. A few said a company has the right to guard its brand and terms, but that chorus was drowned out by devs warning this is how you lose community trust fast.

Key Points

  • The PR removes Anthropic-branded elements from OpenCode per legal requirements, including a system prompt file and a built-in Anthropic auth plugin.
  • The claude-code-20250219 flag is dropped from the anthropic-beta request header; other beta flags (interleaved-thinking, fine-grained-tool-streaming) remain.
  • Provider login UI and documentation are updated to eliminate Anthropic OAuth/Pro-Max flows and hints.
  • A refactor unintentionally drops the User-Agent: opencode/${VERSION} header for non-Anthropic providers (OpenAI, Google, Azure), flagged as a potential regression.
  • Merge is advised after confirming whether the User-Agent header removal is intentional and fixing stale documentation referencing Claude Pro/Max.

Hottest takes

“Anthropic really on a roll making friends in Open Source.” — mellosouls
“they are interested in suing and not buying...” — adampunk
“Hopefully it doesn't come to that with Anthropic.” — jitl
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.