A high-performance web reading layer that converts public URLs into structured, agent-ready intelligence via a single API call.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install project-ghost
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 project-ghost 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 Project Ghost?
Project Ghost serves as a sophisticated web reading layer designed specifically for the needs of AI agents. Rather than feeding raw, noisy HTML into a Large Language Model, this skill distills any public URL into structured intelligence. It provides critical data points such as business intent, named entities, and confidence scores, making it an essential component of the Openclaw Skills ecosystem for developers building research or monitoring tools.
By leveraging Project Ghost, developers can achieve an average of 80% token reduction compared to raw HTML processing. This efficiency not only lowers costs but also increases the accuracy of agentic workflows by filtering out fluff and focusing on high-signal content. As a premier entry in the Openclaw Skills directory, it bridges the gap between messy web data and clean, actionable insights.
Project Ghost Use Cases
- Research agents performing deep dives into company websites or product documentation.
- News monitoring agents tracking specific topics across Hacker News or industry blogs.
- Sales intelligence agents detecting buying signals and business intent from corporate landing pages.
- Legal and compliance agents extracting policy terms and named entities from service agreements.
- Competitive analysis tools requiring structured data from competitor sites like Stripe or OpenAI.
How Project Ghost Works
- The AI agent provides a public URL to the Project Ghost distillation endpoint.
- The system fetches the web content and bypasses standard layout noise to identify core text.
- A specialized extraction layer analyzes the content for business intent, entities, and priority signals.
- The data is validated through an integrity layer to assign a confidence score (0-1).
- The skill returns a structured JSON object containing distilled intelligence and a token savings report.
Project Ghost Setup
To integrate this skill into your workflow, first obtain a free API key from the Project Ghost homepage. Configure your environment as follows:
# Set your environment variable
export GHOST_API_KEY='your_ghost_sk_here'
# Test the distillation endpoint
curl -X POST https://project-ghost-production.up.railway.app/distill \
-H "Authorization: Bearer $GHOST_API_KEY" \
-H "Content-Type: application/json" \
-d '{"url": "https://apple.com"}'
Project Ghost Data Schema & Taxonomy
Project Ghost organizes web data into a refined taxonomy designed for programmatic consumption:
| Field | Description | Type |
|---|---|---|
business_intent |
1-2 sentence summary of page purpose | String |
priority_score |
Relevance signal on a scale of 1-10 | Integer |
entities |
List of companies, people, or products mentioned | Array |
confidence_score |
Reliability metric of the extraction (0-1) | Float |
tokens_saved |
Percentage reduction vs raw HTML | Percentage |
is_high_integrity |
Boolean flag for data reliability | Boolean |
Project Ghost Advanced Features
- High-efficiency token reduction (80%+) for cost-optimized agentic reasoning.
- Integrated integrity layer to filter out low-confidence extractions automatically.
- Multi-agent support for parallel research tasks across thousands of public domains.
- Priority scoring to help agents rank the relevance of multiple crawled sources.
- Support for a wide range of top-tier domains including GitHub, Wikipedia, and Stripe.
SKILL.md
Loading
Related Openclaw Skills
CamScanner
A specialized skill for extracting and organizing public product features, pricing structures, and help documentation from CamScanner to streamline office collaboration.
Deep Research Pro
A professional-grade research framework that automates systematic academic, market, and competitive analysis with mandatory quality gates.
China Eastern Hot Trend
A specialized tool for extracting and organizing public flight schedules, check-in information, and official notices from China Eastern Airlines.
Fantasy Westward Journey Game Intelligence
An automated information retrieval skill for Fantasy Westward Journey that extracts game updates, event details, and official community rules.
Git Commit Guard
A protective git-first workflow skill for AI agents that ensures repository integrity through rigorous pre-work inspection, multi-stage validation, and structured commits.
Deep Research Pro (v2.2 - True Depth)
Deep Research Pro is a rigorous AI research skill that enforces mandatory full-text parsing, structured data extraction, and source verification to ensure maximum analytical depth.
Featured*
