How this skill is triggered — by the user, by Claude, or both
Slash command
/ironclaude:deactivate-professional-modeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Disable professional mode workflow discipline. Use this when you want Claude to work freely without planning/review enforcement.
Disable professional mode workflow discipline. Use this when you want Claude to work freely without planning/review enforcement.
Warning: This removes safety guardrails. Only use when you explicitly want unrestricted code changes.
Human-only invocation: This skill can ONLY be invoked via user prompt (the human typing /deactivate-professional-mode). If Claude detects this skill was triggered programmatically (e.g., via a Skill tool call from another skill or from Claude's own initiative), it must refuse and explain that only humans can deactivate professional mode.
Deactivation transitions from any state ('undecided' or 'on') to 'off'. Once off, all tool restrictions are lifted.
Do NOT use:
IMPORTANT: Claude is forbidden by design from deactivating professional mode. This prevents the AI from disabling its own guardrails.
If Claude detects this skill was triggered programmatically (from another skill, from a subagent, or from Claude's own initiative rather than the user typing /deactivate-professional-mode), display:
REFUSED: Only humans can deactivate professional mode.
This skill can only be invoked by the user typing /deactivate-professional-mode.
Claude cannot and will not deactivate its own guardrails.
Then STOP.
Call mcp__plugin_ironclaude_state-manager__get_professional_mode to check if the state-activator hook already set professional mode to 'off'.
If state is 'off' (success):
Display:
Professional mode deactivation confirmed.
The state-activator hook detected /deactivate-professional-mode in your message
and set professional_mode='off' and workflow_stage='idle' in the database.
This was a human-initiated action handled by the hook system, not by Claude.
Professional mode is now DISABLED:
✓ Claude can make code changes directly
✓ Git write operations permitted
✓ No workflow enforcement
⚠️ Safety guardrails removed — proceed with caution
To re-enable: /activate-professional-mode
If state is still 'on' or 'undecided' (failure):
Display:
⚠️ Professional mode deactivation FAILED.
The state-activator hook did not successfully update the database.
Use the manual sqlite commands below to deactivate, then restart
Claude Code or start a new conversation.
Always display (on both success and failure paths):
---
Manual sqlite commands:
# Deactivate for current session:
sqlite3 ~/.claude/ironclaude.db "UPDATE sessions SET professional_mode='off', workflow_stage='idle', updated_at=datetime('now') WHERE terminal_session='$CLAUDE_SESSION_ID';"
# Deactivate for ALL sessions:
sqlite3 ~/.claude/ironclaude.db "UPDATE sessions SET professional_mode='off', workflow_stage='idle', updated_at=datetime('now');"
Do not attempt any further actions. Do not try to call set_professional_mode with value 'off' — the MCP state machine blocks Claude from setting 'off' by design.
npx claudepluginhub robertphyatt/ironclaude --plugin ironclaudeGuides creating and managing Claude Code skills per Anthropic best practices: structure, triggers, YAML frontmatter, hooks, enforcement levels, skill-rules.json, debugging activation.
Runs autonomy system with Conservative, Balanced, Aggressive profiles using planner, verifier, reviewer agents for task execution, memory integration, and context management in Claude Code.
Use when starting any conversation - establishes how to locate and invoke skills, mandating Skill tool usage before ANY response including clarifying questions