Claude Code Coding Agent

Scope

Use the selected launcher's --version and --help to choose supported model/effort settings. Keep wrappers opaque: their source and environment may contain credentials. Honor explicit choices; compare requested settings with the receipt's effective settings, leaving absent values unverified.

Fix the workspace, permitted actions, and acceptance evidence. Experiments and parallel writers need isolated workspaces. Boundary: before file-only work, strict confinement, or a wrapper/permission conflict, read Boundaries. Permission changes require authorization; the skill grants no extra scope.

Run

Use claude_run.py as the single print-task entry point: