- Applied documentation/security wording cleanups and metadata sync.
- INTERNAL.md Security section: Reworded security/safety language to be neutral and implementation-focused.
- Removed explicit "prompt injection risks" threat framing from the internal documentation text.
Added scripts/persona.py for programmatic persona management:
--list— List all available personas--show <name>— Show persona details--activate <name>— Activate a persona (outputs system prompt, saves state)--current— Show currently active persona--reset— Deactivate/reset to default
Common variations are now supported:
chef→chef-marcodr→dr-medprofessor→professor-stein- And more...
Active persona is saved to ~/.openclaw/persona-state.json and persists across sessions.
Activation history is tracked for easy switching back.
- Persona Count: Documentation now correctly states 20 personas (not 31)
- Path References: Removed hardcoded paths, use relative paths
# List all personas
python3 scripts/persona.py --list
# Activate a persona
python3 scripts/persona.py --activate dev
# Check current
python3 scripts/persona.py --current
# Reset
python3 scripts/persona.py --reset- Privacy cleanup: removed hardcoded paths and personal info from docs