Guides

Answers to the questions people ask before they let a coding agent near production errors. For one stack's setup, see setup by framework; for a specific error message, the error fixes.

• How do I give my coding agent access to production errors?

  Pasted stack traces, shell access to logs, or an error tracker with an MCP server: how to get production errors to Claude Code, Cursor or Codex.

• How to have Claude Code find, fix and resolve production errors

  Claude Code reads production errors over MCP, fixes the code and resolves the error with the commit. Prompts, a claude -p triage run and permissions.

• How do I track errors in an AI agent in production?

  Record each agent failure with the session, tool and action attached, plus breadcrumbs for model and tool calls. Node.js and Python code.

• How do I see errors from my MCP server in production?

  MCP tool errors go back to the client as results and never reach your logs. Wrap each handler and report it, with TypeScript and Python code.