Welcome to buddy Discussions!

[fiorastudio]

👋 Welcome!

We’re using Discussions as a place to connect with other members of our community. We hope that you:

• Ask questions you’re wondering about.
• Share ideas.
• Engage with other community members.
• Welcome others and are open-minded. Remember that this is a community we
  build together 💪.

To get started, comment below with an introduction of yourself and tell us about what you do with this community.

[cleverestx]

Hi, I'm a solo dev just learning the process.

Does this still work with the latest version of Claude CLI/Code?

[fiorastudio]

How it differs from the Anthropic version:

The original /buddy was a system prompt layer baked into Claude Code itself — Anthropic shipped it experimentally, never documented it, and quietly removed it in v2.1.97 (April 9, 2026). It also had a server-side endpoint ("buddy react") that used an Anthropic model to generate reactions to your code. Because both the client feature and the server endpoint lived under Anthropic's control, any update could (and did) kill it.

This version is fundamentally different in architecture:

• MCP server, not a system prompt. It runs as a standalone process that Claude Code (or any MCP client) connects to. Anthropic can't remove it because it doesn't live inside their code.
• No server-side dependency. The original relied on Anthropic's buddy react endpoint for AI-powered reactions. This version handles personality, reactions, and code feedback entirely locally through the MCP server — no external endpoint that can be shut off.
• Portable. Works with Claude Code, Codex, Cursor CLI, Copilot CLI, OpenClaw, Gemini — anything that speaks MCP. The original only worked in Claude Code.
• Persistent state. SQLite database that survives terminal restarts, updates, reinstalls. Your buddy's level, XP, stats, personality — all saved locally.
• Same identity algorithm. If you had a buddy before, it uses the same hash (wyhash + mulberry32, same salt) so you get back your exact species, rarity, and stats from your original accountUuid.
• Way more features. 21 species, 5-stat system, guard mode (structural reasoning analysis), skill coaching, leveling with real progression curves. The original was just a personality overlay with server-side reactions.

Hope this helps!

[fiorastudio]

Hi, I'm a solo dev just learning the process.

Does this still work with the latest version of Claude CLI/Code?

And yes. Latest Claude CLI / code should work just fine.

[fiorastudio]

Hi cleverestx, Thanks for your interest. Yes it does work with latest claude CLI/code.

…

--- *How it differs from the Anthropic version:* The original `/buddy` was a system prompt layer baked into Claude Code itself — Anthropic shipped it experimentally, never documented it, and quietly removed it in v2.1.97 (April 9, 2026). It also had a server-side endpoint ("buddy react") that used an Anthropic model to generate reactions to your code. Because both the client feature and the server endpoint lived under Anthropic's control, any update could (and did) kill it. This version is fundamentally different in architecture: • *MCP server, not a system prompt.* It runs as a standalone process that Claude Code (or any MCP client) connects to. Anthropic can't remove it because it doesn't live inside their code. • *No server-side dependency.* The original relied on Anthropic's buddy react endpoint for AI-powered reactions. This version handles personality, reactions, and code feedback entirely locally through the MCP server — no external endpoint that can be shut off. • *Portable.* Works with Claude Code, Codex, Cursor CLI, Copilot CLI, OpenClaw, Gemini — anything that speaks MCP. The original only worked in Claude Code. • *Persistent state.* SQLite database that survives terminal restarts, updates, reinstalls. Your buddy's level, XP, stats, personality — all saved locally. • *Same identity algorithm.* If you had a buddy before, it uses the same hash (wyhash + mulberry32, same salt) so you get back your exact species, rarity, and stats from your original accountUuid. • *Way more features.* 21 species, 5-stat system, guard mode (structural reasoning analysis), skill coaching, leveling with real progression curves. The original was just a personality overlay with server-side reactions. Hope this helps! Thanks, FioraStudio

On Thu, Jul 9, 2026 at 10:59 AM cleverestx ***@***.***> wrote: Hi, I'm a solo dev just learning the process. Does this still work with the latest version of Claude CLI/Code? How does it vary from the original Anthropic-removed version? — Reply to this email directly, view it on GitHub <#81?email_source=notifications&email_token=CA6GKJEMDED4TEXDKMBIKYL5D6XNVA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZVHA3DINZSUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVRTG633UMVZF6Y3MNFRWW#discussioncomment-17586472>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/CA6GKJF5QIHDFEVHJUHZDIT5D6XNVAVCNFSNUABIKJSXA33TNF2G64TZHMYTEMBWGE4DMOJWHE5UI2LTMN2XG43JN5XDWOJZGIYDOMBXUF3AE> . You are receiving this because you authored the thread.Message ID: ***@***.***>