Launches Vibe.
Usage
Installed command:
vibe # Start server if needed and attach the TUI
vibe [options] [message...]
vibe --web [--port 4321]
vibe --bg "prompt"
vibe server start [--foreground]
vibe server restart
vibe server status
vibe server stop
vibe new # Create and attach a fresh session. Alias: vibe n
vibe sessions [--all] [--live] [--failed] [--limit n] # Alias: vibe ls
vibe sessions prune --empty
vibe send <session-id> "prompt"
vibe storage migrate
vibe storage status
vibe storage fts status
vibe storage fts rebuild
vibe storage fts optimize
vibe storage checkpoint
vibe storage vacuum
vibe search <query> [--cwd project] [--role user|assistant|tool] [--include-tools]
vibe storage search <query> [--cwd project] [--role user|assistant|tool] [--include-tools]
vibe storage import pi <path> [--no-fts] [--rebuild-fts] [--batch-size N]
vibe subagents jobs|active|schedules|status <id>|result <id>|cancel <id>
vibe connect [--ssh|--dist] <target>
vibe gateway telegram --foreground
vibe skill list|show <name>|apis|from-session <session-id> <name>
vibe attach # Alias: vibe a
vibe attach <session-id> # Alias: vibe a <session-id>From a checkout:
mix vibe
mix vibe [options] [message...]Options
--model <provider:model>- Model identifier passed to ReqLLM. Defaults toVIBE_MODELoropenai_codex:gpt-5.5.--api-key <key>- API key for OpenAI-compatible requests.--system-prompt <text>- Override system prompt for direct TUI/--directcalls.--mode <text|json>- Output mode. Defaults totext.--print,-p- Non-interactive mode: process prompt and exit.--bg- Start a background session for the prompt and return immediately.--direct- Use a direct ReqLLM call instead of the tool-capable agent runtime.--stream- Stream direct ReqLLM calls. Default for--direct.--no-stream- Disable direct ReqLLM streaming.--eval <code>- Evaluate Elixir code throughVibe.Eval.--compact- Compact stored trajectory context.--web- Ensure the server is running and open the Phoenix LiveView web console.--port <port>- Port for--web. Defaults to 4321.--keep-recent <n>- Events to keep when compacting. Defaults to12.--checks- Run Vibe validation gates.--codex-usage- Show Codex subscription usage via Codex app-server RPC.--session <id>- Continue or name a persisted session.--sessions- List persisted sessions. Prefervibe sessionsfor server-aware listings.--all- Withvibe sessions, include empty/internal historical sessions.--live- Withvibe sessions, show only live sessions.--failed- Withvibe sessions, show sessions whose preview looks failed.--limit <n>- Withvibe sessions, limit displayed sessions. Defaults to20.--timeout <ms>- Request/eval timeout.--cast <path>- Record the TUI byte stream as a native gzip cast for debugging.--ssh,--dist- Select remote transport forvibe connect.--yes- Accept remote SSH host keys forvibe connect --ssh.--login codex- Sign in with ChatGPT/Codex OAuth.--help,-h- Show this help.--version,-v- Show version.
Examples
vibe
vibe -p "Inspect runtime info with eval"
vibe --model openai_codex:gpt-5.5 "Review this project"
vibe --login codex
vibe --compact --keep-recent 20
vibe --eval "Vibe.OTP.runtime_info()"
vibe --bg "Research this failure in the background"
vibe new --mode json
vibe ls --limit 5
vibe send 20260425-120000-abcd "Use eval to inspect System.version()"
vibe storage search "figma variable"
vibe subagents jobs
vibe connect --ssh localhost:4022 --yes
vibe a
vibe a 20260425-120000-abcd