Human-in-the-Loop Safe Destructive Action Gate

Name: Human-in-the-Loop Safe Destructive Action Gate
Rating: 4.9 (18 reviews)
Author: Anthropic

Verified Secure

Created by Anthropic•agents Workflow

Enforce mandatory interactive approval before agents execute destructive or financial operations.

Supported Runtimes:codexclaude-codecursorwindsurf

Universal 1-Click Installer

npx askills install human-in-the-loop-approval-gate

Installs automatically to ~/.agents/skills or ~/.codex/skills

•Raw Endpoint

Client Trigger & Invocation Simulator

Interactive Agent Cheatsheet

Codex Native Trigger Pattern

Zero-Guessing Syntax

$human-in-the-loop-approval-gate <your task>

Type `$human-in-the-loop-approval-gate` in the chat or invoke via natural language. Codex automatically identifies the skill from ~/.agents/skills or project .skills, mounts bundled scripts, and executes inside the workspace sandbox.

Ready-to-Run Prompt for Codex NativeClick to copy

$human-in-the-loop-approval-gate Display a unified Git diff and require user confirmation before applying multi-file code refactors.

Agent Runtime Simulation (Codex Native)

user@agent:~$ $human-in-the-loop-approval-gate Display a unified Git diff and require use...

Click "Test Trigger Simulation" above to see how Codex Native discovers and runs Human-in-the-Loop Safe Destructive Action Gate in real-time.

Overview & Problem Solved

When modern frontier LLMs (like Claude 3.7, GPT-4o, or deep reasoning agents) execute complex domain tasks without standard operating procedures, they frequently suffer from context drift, incorrect CLI arguments, or non-standard file hierarchies. Human-in-the-Loop Safe Destructive Action Gate eliminates probabilistic guessing by supplying the Agent with an authoritative, sandboxed execution specification.

By packaging verified workflows into a structured SKILL.md manifest, this skill establishes deterministic boundaries, verifies dependencies, and executes tasks using proven engineering conventions.

Operational Impact: Without Skill vs. With Skill

Raw frontier models possess deep knowledge, but probabilistic prompting lacks rigid engineering guardrails. Packaging Human-in-the-Loop Safe Destructive Action Gate into a verified SKILL.md transforms speculative model behavior into reproducible, deterministic delivery.

Without Skill (Raw Probabilistic LLM)

High Failure Risk

Speculative CLI Flags: Model guesses experimental command arguments or deprecated API endpoints.
Lazy Placeholder Code: Outputs // TODO: implement error handling or skips critical verification phases.
Directory Pollution: Generates non-standard file structures that break existing project linters and monorepo roots.
Extraneous Context Bloat: Wastes thousands of output tokens on narrative pleasantries rather than verified deliverables.

// Raw LLM output (Prone to regressions):

I will now write your code. Note that you may need to install some packages...

function process() {
  // TODO: Connect database here
  console.log("Done");
}

⚠️ Unverified: missing types, missing schema validations, no self-healing check.

With Human-in-the-Loop Safe Destructive Action Gate (Deterministic Agent SOP)

Deterministic

Codified Workflow Stages: Follows strict step-by-step algorithms explicitly codified in SKILL.md.
Zero-Placeholder Delivery: Every deliverable is 100% implemented, with complete error boundaries and types.
Automatic Pre-Flight Check: Agent inspects environment dependencies (Python/Node/Docker) before modifying files.
Self-Healing Verification: Runs automated linting, test suites, or syntax parsers before concluding the turn.

// SKILL.md SOP Output (Verified & Sandboxed):

[Stage 1] Pre-flight: schema parsed & dependencies locked.

export async function execute() {
  const validated = schema.parse(input);
  return await auditAndPersist(validated);
}

Passed 100% static typecheck & workspace sandbox rules.

Core Capabilities & Guarantees

Deterministic SOP Enforcement

Follows pre-tested instructions and step-by-step algorithms, avoiding experimental or hallucinated flags.

Multi-Runtime Isolation

Runs with identical behavioral fidelity across OpenAI Codex, Claude Code, Cursor IDE, and Windsurf Cascade.

Sandbox-Scoped Boundaries

Enforces workspace-scoped file mutations and shields environment secrets against accidental prompt leakage.

Self-Healing QA Verification

Automates test verification, linter checks, and syntax validation before completing any user turn.

Agent Execution Pipeline

Intent Matching & Context Ingestion

The agent scans the user request against the skill's YAML frontmatter triggers and loads the verified instructions into active working memory.

Toolchain & Dependency Validation

The runtime verifies that necessary tools, interpreters, and environment configurations exist within the workspace sandbox before proceeding.

Deterministic Execution & Delivery

The agent executes the codified SOP, runs verification checks, and outputs high-fidelity deliverables without extraneous narration.

Validated Prompt Templates

(Click to copy)

“Display a unified Git diff and require user confirmation before applying multi-file code refactors.”

“Prompt the user with explicit details before executing a database DROP TABLE or ALTER migration.”

“Require manual authorization before triggering live cloud deployments or payments.”

Automated Security & Audit Report

Score:98/100 (SAFE)

Filesystem

Workspace Scoped

Cannot escape workspace root

Shell Execution

Instruction Only

Zero system subprocesses

Network Access

Offline Sandbox

No internet egress

API Keys Required

None Required

Runs fully standalone

Enterprise & Production Use Cases

Autonomous Development SprintsEnables coding agents to independently build features, run lint tests, and author PR descriptions that strictly conform to organizational coding rules.

Batch Automation & CI/CD PipelinesAutomates repetitive transformations such as document generation, schema migrations, or test-suite updates inside containerized CI runners.

Standardized Engineering OnboardingDistributes unified development playbooks across all engineering team members using Cursor, Codex, or Claude Code.

Frequently Asked Questions

Q: What makes Human-in-the-Loop Safe Destructive Action Gate different from a standard MCP server?

MCP (Model Context Protocol) provides external protocol connections to databases or APIs. In contrast, Human-in-the-Loop Safe Destructive Action Gate is a high-level operational skill (SKILL.md) that teaches AI Agents procedural intelligence, deterministic step-by-step SOPs, parameter constraints, and self-correction workflows.

Q: Is Human-in-the-Loop Safe Destructive Action Gate compatible with frontier models like Claude 3.7 Sonnet, GPT-4o, and DeepSeek-R1?

Yes. The skill is packaged with standard markdown instructions (SKILL.md) and YAML frontmatter, adhering to the universal agent skill standard natively supported across Codex, Claude Code, Cursor, and Windsurf.

Q: What permissions does Human-in-the-Loop Safe Destructive Action Gate require in agent mode?

Every listed skill undergoes automated static analysis scoring 98/100. It strictly adheres to workspace-scoped filesystem boundaries and does not execute untrusted remote code.

Q: Can our team customize or extend the rules in this SKILL.md?

Yes. Once installed in your project root (.skills/human-in-the-loop-approval-gate/SKILL.md or ~/.agents/skills/human-in-the-loop-approval-gate/), you can customize the instructions to include internal team guidelines, custom linters, or proprietary APIs.

SKILL.md Source Specification

View Raw

---
name: "human-in-the-loop-approval-gate"
description: "Enforce mandatory interactive approval before agents execute destructive or financial operations."
metadata:
author: "Anthropic"
version: "1.0.0"
runtime_compatibility:
- "codex"
- "claude-code"
- "cursor"
- "windsurf"
triggers:
- "require approval for destructive action"
- "add confirmation gate agent"
- "preview diff before execution"
permissions:
fs_read: true
fs_write: "workspace"
shell_exec: false
net_access: false
---

# Human-in-the-Loop Safe Destructive Action Gate

Builds explicit approval checkpoints for high-risk actions (file deletion, database migrations, cloud provisioning, sending emails): structured diff previews, risk classifications, and confirmation prompts.

## When to Invoke
Invoke this skill whenever you need to:
- require approval for destructive action
- add confirmation gate agent
- preview diff before execution

## Execution SOP & Rules
1. Classify tool calls into Read-Only (safe to auto-execute) vs Destructive (requires approval).
2. Display exact arguments and a human-readable summary of the intended side effects.
3. Abort execution immediately if the user rejects the action or cancels the prompt.
4. Log all approval decisions and timestamps in an audit log for compliance tracking.

## Verification Checklist
- Run local linting, syntax validation, and type-checks before reporting completion.
- Ensure all created or modified files adhere to workspace boundaries and avoid uncommitted secrets.
- Verify tests or build commands exit cleanly with return code 0.