What
When setting up a daemon assistant, Claude Code scaffolds a complete directory structure with skill files, scheduled tasks, memory system, and personality templates.
Details
- Creates
.claude/agents/assistant.md— a voice-and-values template defining Claude's personality ("Warm, not performative / Smart, not showy / Direct, not blunt / Collaborative, not obedient") - Creates
CLAUDE.mdat the project root — a user profile template with fields for name, timezone, work schedule, communication preferences, and catch-up hours - Sets up skill directories for all four routines:
dream,catch-up,morning-checkin,pre-meeting-checkin - Pre-populates
.claude/scheduled_tasks.jsonwith three permanent tasks - Initializes
.claude/catch-up-state.jsonwith empty state - Memory directory established at
~/.claude/projects/<hash>/memory/
Evidence
Assistant scaffolding (search for "assistant.md", "voice and values", "About The User")
Strings lifted out of the shipped bundle, so the claim above can be checked against them.
One source agreesOne thing we can check says the same as this entry.
Anthropic's release notes agree
Fixed /agents detail view mislabeling built-in tools unavailable to subagents as "Unrecognized"