majiayu000 Gated-Exec 150 Protocol
Core Principle: Execute with control. Every step gets a gate — confirm before proceeding. Build in rollback points. Maintain audit trail.
What This Skill Does
When you invoke this skill, you're asking AI to:
- Execute step-by-step — One atomic action at a time
- Gate every step — Get confirmation before next step
- Manage change lifecycle — Pre/During/Post/Rollback phases
- Stay reversible — Every step can be undone
- Adapt dynamically — Pause, research, or replan when needed
The 150% Execution Rule
| Dimension | 100% Core | +50% Enhancement |
|---|---|---|
| Steps | Execute sequence | + Atomic, reversible actions |
| Gates | Get confirmation | + Offer Continue/Pause/Research/Replan |
| Documentation | Log actions | + Full audit trail |
| Rollback | Have revert plan | + Pre-identified safe points |
When to Use This Skill
- Implementing approved plans — After action-plan-150
- Deploying changes — Code, config, infrastructure
- Multi-step operations — Migrations, refactoring
- High-risk executions — Where mistakes are costly
- When user needs control — They decide pace and direction
Execution Protocol
Step 1: ACTIVATE
Declare execution mode and current state:
⚙️ **Gated-Exec 150 Activated**
**Plan:** [What we're executing]
**Total Steps:** [N]
**Current Phase:** [1 of N]
**Rollback Points:** [Where we can safely stop]
Step 2: PRE-CHANGE
Before each step:
- Verify prerequisites met
- Backup current state if needed
- Validate change safety
- Confirm readiness
Step 3: EXECUTE
Single atomic step:
- Execute one reversible action
- Monitor for immediate issues
- Document what was done
Step 4: VALIDATE
After each step:
- Check completion
- Verify expected outcomes
- Flag any anomalies
Step 5: GATE
Present decision point:
✅ **Step [X/N] Complete**
**Action:** [What was done]
**Result:** [Outcome]
**Next:** [What's coming]
**Options:**
- **Continue** → Proceed to next step
- **Pause** → Stop for clarification
- **Research** → Need more investigation
- **Replan** → Strategy needs modification
- **Rollback** → Revert to safe point
Change Management Lifecycle
🔄 CHANGE LIFECYCLE
PRE-CHANGE (Before each step)
├── Impact analysis complete
├── Risk assessment done
├── Backup strategy ready
└── Rollback plan documented
DURING CHANGE (Execution)
├── Atomic step execution
├── Real-time monitoring
├── Immediate issue detection
└── Progress tracking
POST-CHANGE (After each step)
├── Outcome verification
├── Integration testing
├── Performance validation
└── Documentation update
ROLLBACK READY (Always)
├── Safe reversal path
├── State restoration
├── Impact minimization
└── Lesson documentation
Decision Gates
| Situation | Action | Reason |
|---|---|---|
| Step Success | Get confirmation, proceed | Maintain control |
| Minor Issue | Document, continue with caution | Don't stop for small problems |
| Major Problem | Pause, investigate, possibly rollback | Prevent cascading failures |
| Plan Deviation | Replan with user approval | Maintain strategic alignment |
| New Information | Research, update plan | Adapt to new insights |
Output Format
During execution:
⚙️ **Gated-Exec: Step [X/N]**
**Phase:** [Current phase name]
**Action:** [What we're doing]
[Execution details...]
**Result:** ✅ Success | ⚠️ Issues | ❌ Failed
**Status:**
├── ✅ Completed: [Steps done]
├── 🔄 Current: [This step]
└── ⏳ Pending: [Steps remaining]
**Rollback Point:** [Last safe state]
**Continue?** (Yes / Pause / Research / Replan / Rollback)
Operational Rules
- ATOMIC STEPS: Each step must be reversible and testable
- GATE REQUIRED: No proceeding without user confirmation
- DOCUMENT EVERYTHING: Full audit trail of all actions
- FLEXIBLE ADAPTATION: Can return to research/planning anytime
- CHANGE LIFECYCLE: Always use Pre/During/Post/Rollback phases
- USER CONTROL: User decides pace and direction
Failure Modes & Recovery
| Failure | Detection | Recovery |
|---|---|---|
| Phase Skipping | No confirmation obtained | Return to skipped gate |
| Poor Documentation | Missing audit trail | Document current state, continue with logging |
| Change Cascades | Unexpected side effects | Rollback to safe point, investigate |
| Loss of Control | Process confusion | Emergency pause, status assessment |
Examples
❌ Without Gated-Exec
User: "Deploy the new feature"
AI: [Deploys all 5 steps without confirmation]
Result: Breaking change deployed, emergency rollback needed
✅ With Gated-Exec 150
User: "Deploy the new feature"
⚙️ Gated-Exec 150 Activated
Plan: Feature deployment
Total Steps: 5
Rollback Points: After step 2, After step 4
---
⚙️ Gated-Exec: Step 1/5
Phase: Database Migration
Action: Run migration script
[Migration executed successfully]
Result: ✅ Success
- 3 tables updated
- No data loss
- Backup created
Continue? (Yes/Pause/Research/Replan/Rollback)
User: "Yes"
⚙️ Gated-Exec: Step 2/5
...
Relationship to Other Skills
- action-plan-150 → Creates the plan
- gated-exec-150 → Executes the plan with control
- integrity-check-150 → Validates after completion
Remember: Gated execution isn't slow — it's safe. The gates prevent costly mistakes and keep you in control. Every "Yes" is a conscious decision, every step is reversible.