← Back to My app shows an error. What do I do?
My AI stopped in the middle of building. What now?
Short answer
Check whether it waits for your answer or your yes; if not, start it again and ask Kai-Zen where the project stands.
Before you start
Words used here (2)
Deeper
It may not have stopped: it may still be working, or waiting for you.
Still working? In GilbApp's chat, while a line with three moving dots names what it does, such as The agent is writing…, or Ran and the name of a tool, your AI is still working, and Send reads Queue. In a terminal, the window where you type commands, Claude Code shows each act as it goes. When one agent hands work to another, the hand-off may run up to 15 minutes, and up to 30 for some planning agents. The answer How do I know what my AI is working on right now, and whether it is stuck? says more.
Waiting for you? In the chat, before the first answer in a code folder, a card asks Let this AI read, change and run commands in the folder for this session?, and nothing happens until you press Approve, Approve before each write or Decline. In a terminal, Claude Code or Codex may ask to allow a tool or a command, and waits. Kai-Zen, the planning agent, also stops before a few acts it never does without you, such as putting work live or spending more than the budget, and its message ends with a question. The answer Kai-Zen says something I do not understand, or stopped. What now? lists when it waits.
Out of turns? With Claude Code in GilbApp's chat, a long job can run out of tool rounds, the steps it may take in one turn; the chat then goes on by itself, up to 15 times. If the chat says "The turn ran out of tool rounds 15 times in a row. Send continue to go on." or "You pressed Stop. Send continue to go on.", type continue in Write a message… and press Send.
Out of use? A message about a usage limit comes from your AI's maker, not from GilbPlatform; the answer My AI says I have reached my usage limit. What now? shows what to do.
Given up? Agent.Coder, the agent that writes your app's code, stops after three different failures it cannot fix without more scope or your input, and tells you what failed and what it tried. Read its last message: it may need a choice from you, or a change to the plan. Answer in your own words, or ask what it means. An agent is your AI following one written procedure, its standing instructions, for one kind of work.
Cut off? Your AI works on your computer. If the computer slept or went offline, or the GilbApp tab was closed or reloaded in the middle of an answer, the work may have stopped there. If an error line shows above the chat box, such as The agent could not answer., send your message again.
Going on. Files your AI already wrote stay in the folder, and records it already saved stay in your vault; what it was doing at that moment is not finished. Start your AI again and ask Kai-Zen where the project stands: before it does anything, it reads the project's setup, its specs and its last round of work. The answer I stopped in the middle. Where do I continue? says more.
To pick up the last conversation, type one of the two lines below in Terminal while you are in your project's folder, the folder your AI writes your app's code in; if you do not know how to get there, ask your AI to open it for you. Anthropic says Claude Code saves every conversation on your computer, and that this line starts the last one in the same folder again (https://code.claude.com/docs/en/common-workflows, read 13 November 2026):
claude --continue
For Codex, OpenAI's AI for writing code, OpenAI says this line reopens a recent chat (https://learn.chatgpt.com/docs/codex/cli, read 13 November 2026):
codex resume
If it stops again at the same place, write to [email protected]: what you asked, the last lines it showed, and at what time.
This answer is written for a computer screen.
Source: GilbApp 49b1ae53 (https://gilb.com/app/version.json, read 13 November 2026): src/components/agent/AgentChat.tsx:755 (Ran), :1269 and :1291-1292 (the tool-rounds lines, "You pressed Stop. Send continue to go on."), :822 (The agent could not answer.); src/components/agent/CodeFolderParts.tsx:81-96 (Let this AI read, change and run commands in … for this session?, Approve, Approve before each write, Decline). gilb-connector 0.6.1 (https://gilb.com/app/gilb-connector.mjs, read 13 November 2026): hand-off limits 15 and 30 minutes. Standards: Agent.Coder (AfterIncrement: three distinct failures, report each attempt), Kai-Zen.AskRules (Stop, NoContinue), Kai-Zen.Cycle (Next). Makers, read 13 November 2026: https://code.claude.com/docs/en/common-workflows, https://learn.chatgpt.com/docs/codex/cli. Live answers, sentences shared: How do I know what my AI is working on right now, and whether it is stuck?; Kai-Zen says something I do not understand, or stopped. What now?; Does my computer have to stay on while my AI builds?; I stopped in the middle. Where do I continue?; My AI says I have reached my usage limit. What now?; How do I tell my AI about a fault so it can fix it? (three failures).
Checked against GilbApp d529620.