claude-code-settings

Claude Code memory and agent behavior are configured through CLAUDE.md files and .claude/agents directories to define project-specific instructions and agent directives. This centralizes memory hierarchy definitions and output style controls for tailored team workflows.

The Claude Code memory hierarchy uses CLAUDE.md files to define global and project-specific configuration rules. It centralizes memory management by layering instructions from managed settings down to local project scopes for consistent agent behavior.

Yes, Claude Code supports per-project model selection and output style configuration through settings.json scopes. You can tailor model behavior at managed, user, project, and local levels to fit specific code-centric workflows.

Claude Code governance enforces the presence of SKILL.md frontmatter containing name and description fields. It supports optional scripts, references, and assets to define the skill's deployment boundaries and governance structure.

Claude Code settings span managed, user, project, and local scopes, with managed settings enforcing governance boundaries while local settings allow individual project overrides. Configuration hierarchy dictates that higher scopes constrain lower ones.

Organize .claude/agents configuration by defining agent directives within the directory structure, then applying per-project settings to tailor code-centric workflows. This ensures explicit model selection and output style controls across team deployments.