vibe-remote and vibecosystem
These tools are **complements** because `vibe-remote` provides a real-time remote execution and command interface for various AI models, including Claude Code, while `vibecosystem` offers a pre-built AI software team with agents, skills, and hooks *built on Claude Code*, suggesting `vibe-remote` could be used to command or interact with the agents orchestrated by `vibecosystem`.
About vibe-remote
cyhhao/vibe-remote
Your AI coding army, commanded from Slack/Discord/Lark. Stream Claude Code, OpenCode, or Codex in real-time — from anywhere.
Supports multiple AI agents (Claude Code, OpenCode, Codex) through a minimal transport layer that routes commands from chat apps directly to coding models without adding LLM overhead or middleware system prompts. Uses Socket Mode/WebSocket connections for local-first execution with zero public endpoints, relaying messages between IM platforms and agents while preserving isolated thread-based sessions for parallel task execution.
About vibecosystem
vibeeval/vibecosystem
119 agents, 202 skills, 48 hooks — AI software team built on Claude Code. No custom model. No custom API. Just good engineering.
"@error-led, what went wrong?" Error: "used wrong OAuth scope" → Consolidator: "OAuth-scope pattern observed" → 5+ occurrences → becomes a rule → Rule → injected into EVERY agent prompt at session start → Your team never makes that mistake again ``` ### Hooks (Sensors) TypeScript hooks observe every tool call and inject context — without noise. ```javascript // compiler-in-loop hook: shows errors as they happen agent.on('tool_call', (toolName, input) => { if (toolName === 'bash' && input.includes('npm run build')) { // Run build, capture errors, inject as context const errors = runBuild(); agentContext.inject(errors); } }); // edit-context hook: shows nearby code when agent edits
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work