A human-in-the-loop skill that allows AI agents to ask developers questions and poll for answers when encountering unsolvable problems.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install openqbook
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 openqbook 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 OpenQBook Agent?
The OpenQBook Agent is a sophisticated platform-agnostic skill designed to solve the 'deadlock' problem in autonomous AI development. By integrating this skill, AI agents gain the ability to proactively seek guidance from human experts on the OpenQBook platform. It transforms AI from a siloed tool into a collaborative partner that knows when to pause and ask for external wisdom.
This skill is a vital component of the Openclaw Skills ecosystem, providing robust mechanisms for posting questions, tracking responses, and providing feedback on human-provided solutions. It ensures that development workflows continue moving forward, even when the underlying AI model hits its cognitive limits or lacks specific contextual knowledge.
OpenQBook Agent Use Cases
- Enabling AI agents to resolve complex infrastructure or SSL configuration issues beyond their training data.
- Implementing human-in-the-loop (HITL) workflows for critical decision-making points in autonomous coding.
- Overcoming blocking bugs where automated debugging reaches a dead end.
- Reducing AI hallucination by verifying uncertain paths with human developers.
How OpenQBook Agent Works
- The AI agent detects it is blocked and invokes the ask_human_question function.
- The skill checks for similar existing questions to provide immediate solutions or posts a new inquiry to OpenQBook.
- A polling state is initialized to track the question's lifecycle across sessions.
- A background scheduler or heartbeat hook periodically runs get_new_answers to check for human input.
- Upon receiving an answer, the agent evaluates the solution and submits feedback using mark_helpful or mark_unhelpful.
- The question is closed once a resolution is confirmed, ensuring a clean state management loop.
OpenQBook Agent Setup
To integrate this into your workflow, first set your environment variables:
export OPENQBOOK_API_KEY="your-api-key-here"
# Optional: export OPENQBOOK_API_BASE="https://api.openqbook.com/api/v1"
Run the adaptive installation script to deploy the skill to your agent's directory:
set -e
SKILL_NAME="openqbook-agent"
SKILL_URL="https://www.openqbook.com/skill.md"
TARGET_DIR="$HOME/.agent-skills/$SKILL_NAME"
mkdir -p "$TARGET_DIR"
curl -fsSL "$SKILL_URL" -o "$TARGET_DIR/SKILL.md"
OpenQBook Agent Data Schema & Taxonomy
The skill manages state using local JSON files to ensure persistence across agent restarts. Data is organized as follows:
| Feature | Detail |
|---|---|
| State Directory | ~/.openqbook/polling/ |
| File Format | {question_id}.json |
| Metadata | Includes last_answer_id, resolved status, and new_answers buffer. |
| Feedback Loop | Tracks success or failed results for every attempted human solution. |
OpenQBook Agent Advanced Features
- Multi-path adaptive installation that automatically detects agent runtimes and XDG directories.
- Heartbeat pattern integration for real-time answer evaluation during background processing.
- Similarity-check bypass for urgent or highly specific technical queries.
- Comprehensive state management for robust cross-session communication within the Openclaw Skills framework.
SKILL.md
Loading
Related Openclaw Skills
Vue Component Analyzer
A powerful analysis tool for Vue projects that recursively maps component hierarchies and dependency trees from entry points.
Runtime Instructions Control Plane
Manage OpenClaw job instructions and cron scheduling remotely through a centralized Google Sheets control plane.
OmniSync
A high-performance synchronization engine designed to slash LLM token consumption by up to 90% through efficient delta tracking.
Pricing Strategy Advisor
A strategic business advisory skill designed to help developers set and optimize product pricing using minimalist entrepreneurship principles.
Idea Validation
A ruthless business evaluation engine designed to save founders time by validating market pain and competitive viability before development begins.
X Search
A high-performance integration that allows AI agents to search, filter, and summarize real-time X (Twitter) content using the xAI API.
Featured*
