▶ quest log
Claude Dynamic Workflows: orchestrating parallel verifying subagents in Claude Code
◇ Could this help me?
Yes. This is a real, official Claude Code feature that orchestrates many parallel subagents on large tasks with built-in verification, and it is genuinely useful for big codebase audits, large migrations, and critical validation where independent checking matters. It is a general Claude Code capability rather than a project-specific change: use the built-in feature and official documentation instead of any third-party DM-gated "skill", and mind the higher token cost by testing on a small scoped task first.
Create a dynamic workflow to audit the ./src/auth directory for security and performance issues. Plan the work, run parallel subagents that each inspect the code from an independent angle, have other subagents challenge and verify their findings, and iterate until the results converge into a single prioritized list with file and line references. Run it on this one scoped directory first so I can see the token cost before applying it more broadly, and pause for my confirmation before executing.