메뉴
Troubleshoot why the agent stopped replying to iMessages, in priority order.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | diagnose-no-imessage-replies |
| description | Troubleshoot why the agent stopped replying to iMessages, in priority order. |
Work through these in order and stop at the cause. Report the cause and the specific fix.
Budget cap. Check the daily budget. When the cap is hit the agent goes SILENT with no error (texts just stop). Raise the limit or wait for reset.
Bridge alive + ingesting. Is the iMessage node still feeding messages in? Look for recent inbound items (source "import", tagged imessage) via recall/memory. If the newest is hours/days old, the bridge process on the node is down — have the user restart it (launchctl kickstart -k …) and confirm Messages.app is running and signed into iMessage.
Text decoding. Are recent captured messages real text, or a replacement char (�)? If they decode to �, the attributedBody decoder is failing on that node's macOS, so the trigger word never matches. Update + restart that node's bridge.
Trigger + allowlist. A reply requires the trigger word (whole-word match) from an allowed sender or chat (unless the bridge runs with --respond all). Confirm the message actually contained the trigger and the sender/chat is on the allowlist.
Node asleep. A sleeping Mac stalls Messages sync and osascript sends. IMPORTANT: silent mode / Do Not Disturb does NOT affect the bridge — it reads chat.db and sends via Messages regardless of notification settings. Only the machine SLEEPING breaks it; keep the node awake.
Most of these require the user to act on the node itself (you usually can't SSH it) — give them the exact step.
User asked: {{input}}
Guide configuring proactive outreach so the main reaches out (digest + live decisions) and the Mac app surfaces it with notifications, an overlay list, and inline buttons.
Roll out new openAGI code to the main and to Mac nodes (iMessage / computer-use), and restart them safely.
Troubleshoot opaque model/tool failures and runaway budget — and their usual root causes.
Connect an MCP server (catalog or custom URL/stdio) and make its tools usable — including completing OAuth on a headless host.
Operate a connected computer-use node to accomplish an on-screen task — screenshot, reason, act in a loop.
Set up a recurring check that only does work (and spends tokens) when there's something new — e.g. "every hour, check BuildBetter for new calls".
