confirm-form

What This Skill Does

The confirm-form skill acts as an intermediary layer between the AI agent and the human user for complex decision-making processes. When an agent identifies that a task requires user validation, consensus, or a choice between multiple technical paths, it utilizes this skill to structure the information into a professional HTML form. By generating a Gist-hosted form, the AI ensures that technical context, research findings, and reasoning are presented clearly, preventing 'context loss' that often happens in simple chat threads. The skill also facilitates structured data collection by parsing the user's responses back into a machine-readable format, which the agent can then use to proceed with the next steps of a project.

Installation

To add this capability to your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/xiaozhuang0127/confirm-form

Ensure that you have Node.js installed on your system, as the skill relies on a local generation script to process the JSON-based forms.

Use Cases

• Complex Refactoring: When the agent finds multiple ways to optimize a code base and needs the developer to weigh in on trade-offs.
• Batch Approval: When the agent has completed multiple research sub-tasks and needs a single sign-off or individual corrections for each.
• Uncertainty Resolution: When the agent hits a logical impasse where the 'best' path depends on user preference or proprietary business context.
• Requirements Gathering: Using the form as a structured intake method to define project parameters before the agent begins work.

Example Prompts

1. "I've finished the initial analysis of the database migration. Can you create a confirm-form that presents the three performance trade-offs I found and asks the user to pick one?"
2. "Generate a confirmation form based on the findings in our recent market research project so the stakeholder can review my recommendations."
3. "Create a form for the user to review the batch of bug fixes I've proposed, allowing them to approve or request changes for each one individually."

Tips & Limitations

• Context is King: Always provide raw data in the findings section. Do not just summarize; give the user enough info to verify your logic.
• Recommendation Tagging: Always use 【我的推荐】 in the basis field of your preferred option to clearly signal your best judgment to the user.
• Limit Scope: While the form is flexible, keep individual forms under 10 questions to ensure high completion rates.
• File Management: Ensure you follow the file naming convention for archives (YYYY-MM-DD_<formId>.json) to maintain a clear audit trail of all project decisions.