A repo-aware questioning protocol for OpenClaw that ensures task clarity and technical readiness before an AI agent begins implementation.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install asking-until-100
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 asking-until-100 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 Asking Until 100?
Asking Until 100 is a sophisticated questioning framework designed for OpenClaw agents to mitigate risks in complex coding and architectural tasks. By slowing down execution when tasks are underspecified, this skill forces the agent to reach a target readiness level through structured clarification rather than making faulty assumptions. It is particularly valuable for high-stakes projects where getting the implementation right the first time is critical.
This skill leverages repository awareness to ensure that questions are high-signal, avoiding redundant inquiries about facts already present in the codebase. By integrating with Openclaw Skills, developers can customize the intensity of the questioning process based on the specific needs of their project or the complexity of the task at hand, moving from basic assumptions to high-rigor reports.
Asking Until 100 Use Cases
- When project requirements or success criteria are ambiguous or underspecified.
- High-stakes architectural changes where errors are expensive to revert.
- Complex debugging tasks requiring deep repository context and hypothesis validation.
- Initial project discovery phases that require building out a structured question report.
- Ensuring execution rigor for build-related tasks with critical dependencies.
How Asking Until 100 Works
- The agent loads the core instructions and any repo-local configuration, such as .asking-until-100.yaml.
- It classifies the current task into categories like coding, architecture, or debugging to determine the required rigor.
- The skill inspects the repository to gather existing facts, ensuring questions are high-value and not redundant.
- It estimates task readiness based on configured dimensions and selects an appropriate questioning mode (fast, guided, deep, or report).
- The agent asks the most impactful questions or generates a comprehensive report before taking any irreversible actions.
- It enforces an execution gate, blocking action for high-rigor tasks until sufficient clarification is provided.
Asking Until 100 Setup
To begin using this protocol with Openclaw Skills, ensure you have the necessary scripts and configuration files in place within your environment.
- Initialize your project configuration by copying a profile:
cp assets/default-profile.yaml .asking-until-100.yaml
- Validate your configuration profile using the provided utility:
python scripts/validate_config.py .asking-until-100.yaml
- For a preview of how the questioning output will look for a specific prompt:
python scripts/preview_question_report.py --prompt "Your task description here"
Asking Until 100 Data Schema & Taxonomy
The skill uses a structured data approach to manage readiness and reporting. It relies on internal references to maintain consistency.
| Component | Description | Format |
|---|---|---|
| Config File | Stores profiles, intensity settings, and precedence. | .asking-until-100.yaml |
| High-Rigor Report | Detailed document containing hypothesis and structure. | Markdown |
| Question Patterns | Predefined templates for high-quality inquiry. | question-patterns.md |
| Project Structure | A deterministic rendering of the proposed codebase. | Python Script Output |
| Readiness Dimensions | Metadata for evaluating if a task is ready to proceed. | protocol.md |
Asking Until 100 Advanced Features
- Repo-local configuration merging for granular control over agent behavior at the directory level.
- Multiple questioning modes ranging from fast for simple tasks to report for complex builds.
- Deterministic project structure rendering to visualize changes before execution begins.
- Integrated execution gates that block implementation until specific readiness criteria are met.
- Specialized profiles optimized for advanced reasoning models within Openclaw Skills to handle high-ambiguity requests.
SKILL.md
Loading
Related Openclaw Skills
SearXNG Tool for OpenClaw
A privacy-focused plugin that enables OpenClaw agents to perform web searches via a self-hosted SearXNG instance without relying on paid search APIs.
GDPR Compliance Tracker
A professional-grade assessment tool that generates prioritized GDPR remediation steps and compliance scoring via a certified security API.
Deep Dialogue
Deep Dialogue is a structured framework for exploring complex philosophical, existential, and psychological issues through systematic Socratic and dialectical inquiry.
Turing Tavily Web Search
A powerful web search skill for AI agents that leverages the Turing Tavily proxy to fetch real-time information and extracted web content.
HD AIGC Skills
A comprehensive AIGC client for generating high-resolution AI images and videos using models like Seedream and Kling within the Openclaw ecosystem.
Eastmoney Financial Search (Miaoxiang)
A professional financial intelligence search skill leveraging Eastmoney's Miaoxiang engine to provide authoritative, real-time market data and analysis.
Featured*
