A specialized safety and verification framework for AI agents to handle sensitive operations and elevated permissions with explicit user consent.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install optional-strict-instructions
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install optional-strict-instructions using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Optional Strict Instructions?
The Optional Strict Instructions skill provides a robust protocol for AI agents to manage high-risk tasks such as system modifications, sudo execution, and permanent file deletions. By prioritizing user choice over automation efficiency, this Openclaw Skills component ensures that every sensitive operation is preceded by resource verification and a clear presentation of execution options. It effectively eliminates the risk of unauthorized automated changes by establishing a strict human-in-the-loop requirement for operations involving elevated permissions or critical system components.
This skill is essential for developers building agents that interact directly with the OS or manage sensitive data. It transforms potentially destructive commands into a structured dialogue, allowing users to select between various safety levels, such as dry runs, backups, or direct execution. By integrating this skill, your agent adheres to a professional standard of transparency and security, ensuring that instructions are followed precisely without unauthorized substitution of methods.
Optional Strict Instructions Use Cases
- Managing operations that require sudo or elevated system permissions.
- Safeguarding file deletion or modification requests to prevent accidental data loss.
- Handling complex system changes where multiple technical approaches (e.g., truncate vs. delete) are available.
- Verifying resource existence and ownership before presenting execution paths to the user.
- Implementing a mandatory confirmation pause for any sensitive or irreversible task.
How Optional Strict Instructions Works
- Verification: The skill first checks for the existence of the target resource and gathers metadata like size, permissions, and ownership.
- Option Presentation: It generates a structured list of choices, including safety-first options (like backups or moves to trash) and privileged options (like sudo commands).
- Explicit Wait: The agent pauses execution and waits for an unambiguous numerical or text-based choice from the user.
- Strict Execution: Once confirmed, the command is executed exactly as described in the chosen option without attempting unauthorized alternatives.
- Reporting: The skill verifies the operation outcome and provides a detailed report, stopping immediately if authentication or permissions fail.
Optional Strict Instructions Setup
To deploy this skill within your Openclaw Skills environment, ensure your agent has the necessary permissions to read file metadata and interact with the terminal.
# Clone the skill repository into your project
git clone https://github.com/openclaw/skills.git
cd skills/optional-strict-instructions
Reference the skill in your agent's configuration file to enable the interactive verification prompts for sensitive CLI operations.
Optional Strict Instructions Data Schema & Taxonomy
The skill organizes metadata and user interactions using a structured taxonomy to ensure consistency across different system environments.
| Attribute | Description |
|---|---|
| Resource Context | Captures path, owner, permissions, and timestamps for the target object. |
| Option Matrix | A list of valid execution methods ranked by risk and privilege level. |
| Confirmation State | A boolean or enum tracking whether the user has provided explicit consent. |
| Execution Log | A detailed record of the exact command run and its exit status. |
| Failure Context | Specific error messages for authentication failures or permission denials. |
Optional Strict Instructions Advanced Features
- Multi-stage authentication handling that stops execution immediately upon sudo password failure.
- Built-in safety patterns for reversible operations, such as moving files to a recycle bin instead of permanent removal.
- Support for dry-run modes that show proposed changes before they are committed to the system.
- Intelligent conflict resolution when multiple valid paths exist for a single user request.
- Seamless integration with other Openclaw Skills to provide a unified security layer for multi-agent workflows.
SKILL.md
Loading
Related Openclaw Skills
Laylatul Qadr Guide
A sophisticated spiritual companion designed to maximize worship during the last ten nights of Ramadan through personalized schedules and behavioral science.
Bifrost SLPx Info
A comprehensive tool for AI agents to query Bifrost SLPx liquid staking protocol data, exchange rates, and user balances across multiple EVM chains.
FAERS Multi-Drug SOC Planner
An automated research architect for creating comparative pharmacovigilance study designs using FAERS data within specific MedDRA System Organ Classes.
Baidu Web Search (Turing Proxy)
A specialized search skill that enables AI agents to query the Baidu search engine via the Turing proxy for real-time Chinese web data.
Breakfast Recommender
An intelligent AI assistant that manages fridge inventory and provides nutritious, seasonal breakfast recommendations based on your available ingredients.
Resource Master
Resource Master is a professional assessment tool designed to calculate and consolidate server hardware requirements based on specific product knowledge bases.
Featured*
