Guard Destructive Actions
Require confirmation before irreversible operations.
Guard Destructive Actions is a free MCP Academy lesson on CoddyKit — lesson 4 of 4. You can read the complete lesson below for free — then practise it hands-on in the browser with a built-in code editor and a 24/7 AI tutor. It is part of the MCP Academy learning path, one of 4 lessons in the course, and your progress syncs across the web and the CoddyKit app.
Some Actions Cannot Be Undone
Deleting records, sending emails, charging cards, and pushing code are irreversible operations. These deserve more caution than a harmless read. ⚠️
Reversibility Is the Test
Ask one question of each tool: can I easily undo this? If the answer is no, treat it as high risk and wrap it in extra safeguards before it can run.
Keep a Human in the Loop
For risky actions, pause and ask a person to approve. Human-in-the-loop approval means an injected instruction cannot quietly trigger real damage.
Confirm Before You Commit
A good pattern is to preview the change and require an explicit confirm flag. This tool refuses to act until the caller has clearly opted in.
@mcp.tool()
def delete_user(id: int, confirm: bool = False) -> str:
if not confirm:
return "Set confirm=true to delete."
return remove(id)Show What Will Happen
Before a destructive call, return a clear summary of the exact effect, like which rows vanish. A precise preview lets the approver catch a mistake in time.
Prefer Soft Over Hard
When you can, make actions recoverable: a soft delete that marks a row instead of erasing it lets you reverse a bad call without losing the data.
Make Repeats Safe
Design destructive tools to be idempotent where possible, so calling delete twice does no extra harm. Safe repeats reduce the cost of a confused retry.
Limit Scope and Volume
Cap how much one call can affect. A delete tool that can only remove a single id by id is far safer than one that can wipe a table in one shot.
Use the Sampling Approval Flow
MCP lets a server ask the host to confirm with the user before acting. Routing risky steps through user approval keeps the person in control of the outcome.
Log Every Sensitive Action
Record who triggered each destructive call, with what arguments and when. A clear audit trail lets you detect abuse and recover when something goes wrong.
Default to Caution
When you are unsure whether an action is safe, treat it as if it is not. Defaulting to asking first costs a moment and prevents irreversible mistakes.
Quick Check
Which approach best guards an irreversible delete?
Recap
Treat irreversible actions specially: require confirmation, preview effects, prefer soft deletes, limit scope, and log everything. Default to caution. ✅
Frequently asked questions
Is the “Guard Destructive Actions” lesson free?
Yes — the full text of “Guard Destructive Actions” is free to read here on the web, and the MCP Academy course includes 4 lessons in total. To practise it interactively (a built-in code editor and a 24/7 AI tutor) and unlock the rest of the MCP Academy course, upgrade to CoddyKit PRO.
What will I learn in “Guard Destructive Actions”?
Require confirmation before irreversible operations. You practise MCP Academy with hands-on code you run directly in the browser, and a 24/7 AI tutor answers your questions as you work through the lesson.
Do I need any experience to start MCP Academy?
No prior experience is required. MCP Academy on CoddyKit is structured for beginners through advanced learners; this is — lesson 4 of 4, so you can start here or from the beginning and move at your own pace.
How long does the “Guard Destructive Actions” lesson take?
Most CoddyKit lessons take about 5–10 minutes. Each one is bite-sized and interactive, so you make steady progress and pick up exactly where you left off across the web and the app.
Can I write and run code in this MCP Academy lesson?
Yes. Every MCP Academy lesson includes a built-in code editor, so you write and run real code right in your browser and get instant AI feedback — no local setup required.
All lessons in this course
- Threats Unique to MCP
- Least-Privilege Tool Access
- Validate & Sanitize Everything
- Guard Destructive Actions