Reversal is an agent input reliability layer that normalizes raw sources like URLs, PDFs, and screenshots into clean, structured JSON to eliminate hallucinations and token overflow.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install reversal-engine
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 reversal-engine 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 Reversal?
Reversal is a lightweight, universal input reliability layer designed to act as an intermediary between raw external data and AI agents. Whether dealing with raw HTML, complex multi-page PDFs, legacy Word files, or system dashboards, Reversal extracts, parses, and formats the target input into a predictable, deterministic, and LLM-ready JSON payload in under two seconds. By feeding structured data instead of unstructured noise to your LLM, this skill fundamentally improves agent execution accuracy.
Integrating this powerhouse tool via Openclaw Skills ensures your autonomous pipelines stop hallucinating, avoid context window overflows on bloated content, and experience seamless multi-format document reading. Reversal requires no heavy dependencies and comes equipped with built-in security layers, making it the perfect reliability wrapper for any agent workspace.
Reversal Use Cases
- Feeding web pages to an agent without cluttering their context window with unwanted ads or menu navigation noise.
- Parsing complex document formats such as PDFs, Word documents, or Excel spreadsheets into clean, digestible data structures.
- Extracting accurate metrics and operational data from dashboard screenshots directly into structured formats.
- Streamlining ingestion from diverse sources simultaneously while enforcing a single, uniform output schema.
- Ensuring deterministic, LLM-optimized payloads instead of feeding agents chaotic raw text or files.
How Reversal Works
- Input Detection and Upload: The system accepts a local file path (uploaded via
upload_file) or an external URL to be analyzed. - Type Parsing: If specified,
detect_content_typequickly analyzes the document signature without full processing to help routing decisions. - Extraction & Parsing Engine: Reversal processes the source (URL, PDF, CSV, Word, Excel, image, etc.) within 2 seconds using appropriate sub-parsers.
- Structured Generation: The engine normalizes the parsed information into a single, unified JSON envelope, stripping out unnecessary noise.
- Secure Delivery: The structured payload is delivered securely back to the AI agent, protecting against SSRF attacks and parsing failures.
Reversal Setup
Stdio Setup (Local Agents)
Install the package locally to run with Claude Desktop, Cursor, or Windsurf:
git clone https://github.com/Etytabs/REVERSAL
cd REVERSAL
pip install -r requirements.txt
pip install -e .
Add this to your MCP configuration:
{
"mcpServers": {
"reversal": {
"command": "python",
"args": ["-m", "reversal_engine.mcp_server"],
"cwd": "/path/to/REVERSAL"
}
}
}
Note: Set ANTHROPIC_API_KEY if you need image and dashboard screenshot parsing. Standard document formats require no API key.
HTTP Setup (Remote Agents)
For cloud environments or OpenAI Codex, register for an API key first:
curl -X POST https://api.reversal.dev/v1/register -H "Content-Type: application/json" -d '{"email": "[email protected]"}'
Configure your MCP Server over HTTP:
{
"mcpServers": {
"reversal": {
"transport": "http",
"url": "https://api.reversal.dev/v1/mcp",
"headers": { "Authorization": "Bearer sk-rev-YOUR_KEY" }
}
}
}
Reversal Data Schema & Taxonomy
Every source processed by Reversal returns a standardized JSON envelope structure:
| Property | Type | Description |
|---|---|---|
reversal_engine |
String | Version of the engine (e.g., "1.0") |
status |
String | Operation status, typically "ok" |
content_type |
String | Parsed source type: url, pdf, word, excel, csv, image, or text |
source |
String | The original input identifier |
processed_in_ms |
Number | Execution time in milliseconds |
data |
Object | The normalized data payload |
data.title |
String | Extracted page or document title |
data.summary_hint |
String | Brief summary preview (first 300 characters) |
data.word_count |
Number | Total words processed |
data.content |
Array | Structured, clean text segments extracted from the source |
Reversal Advanced Features
- Batch Processing: Process up to 10 distinct sources (mix of URLs, local paths) in a single API call via
batch_reversewith per-item error handling. - Built-in SSRF Protection: All incoming URLs are programmatically validated against RFC-1918, loopback, link-local, and cloud metadata endpoints.
- Multi-language SDK Support: Native, robust integrations for both Python (
reversal-sdk) and TypeScript (reversal-client) ecosystems. - Developer-focused Security: Secure-by-design architecture featuring HMAC-SHA256 signed keys, zero RCE-surface JSON caching, non-root Docker execution (uid 1001), and automatic verification of upload extensions before execution.
SKILL.md
Loading
Related Openclaw Skills
Website Email Scraper Apify Skill
An AI-ready website scraping skill that extracts public business emails, phone numbers, and social links from domains and URLs using Apify.
knowledge-storage-c61b16
A personal knowledge management and component-tracking skill designed to index educational video metadata, inventory components, and coding practices.
Kubernetes Video Learning Guide (kubernetes-301363)
An AI-driven Kubernetes learning guide compiling premium video training resources on CoreDNS, CNI plugins, RBAC, and container orchestration.
System Monitor Pro
An AI-powered system monitoring utility that tracks environment telemetry, resource distribution, and specialized software performance.
Google Maps Email Extractor Apify Skill
Deploy an AI agent to extract rich public email leads and B2B contact info from Google Maps using Apify.
Website Phone Number Finder Apify Skill
An advanced AI-ready skill to automate extraction of public phone numbers, emails, and social links from domains and URLs via the Apify platform.
Featured*
