AgentBell: Code Companion

A trusted decision center for Claude Code, Codex, and developer automation. Review the exact request, respond securely, and see what happened next.

AgentBell is a personal trusted decision center for developer automation. Bring bounded requests from Claude Code, Codex, and supported connectors to your iPhone or iPad, review the exact context, and record a decision without exposing a remote terminal.

Approve only what you intend. Every request shows its verified source, action, target, environment, risk, deadline, and what approval will do. After you respond, follow a clear timeline from human decision to executor claim and final result: succeeded, failed, or not yet confirmed.

Features: • One Inbox for Claude Code, Codex, and supported connectors • Bounded confirm, deny, choice, and form responses • Source trust, request provenance, risk, and integrity details • Execution receipts that keep approval separate from success • Activity summaries for Macs, projects, and coding sessions • Live Activities and reliable catch-up when notifications are delayed • Connection health, access review, and revocation • Quiet hours and risk-aware notification preferences • 30-day on-device activity and approval history • One-command Mac integration setup

AgentBell does not start or take over agents, mirror your terminal, send prompts, provide remote keyboard input, or automatically approve requests. It is independently developed and is not affiliated with, endorsed by, sponsored by, or officially connected with Anthropic or OpenAI.