COPPA Guard is a local, pre-release scanner that detects children’s privacy compliance risks in product copy and returns prioritized remediation guidance.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install coppa-guard
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 coppa-guard 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 COPPA Guard?
COPPA Guard is a lightweight compliance guardrail for US-facing children’s apps, games, ecommerce products, and educational services. It scans marketing copy, privacy policies, and app store descriptions before publication for language that may trigger COPPA obligations, including collecting children’s personal information, behavioral advertising, missing verifiable parental consent, third-party disclosure, and persistent identifier tracking.
Designed for Openclaw Skills workflows, the tool runs locally with Python’s standard library, makes no network requests, and produces actionable findings for developers, product teams, and AI agents. It is intended as a fast pre-publication checkpoint rather than a replacement for a complete legal audit or professional legal advice.
COPPA Guard Use Cases
- Pre-screen app store descriptions for children’s privacy triggers before submission.
- Review privacy policies and product copy for under-13 data collection language.
- Detect claims about personalized or behavioral advertising aimed at children.
- Flag statements that imply data collection without verifiable parental consent.
- Identify third-party sharing or sale of children’s data.
- Check for persistent identifiers and cross-site or cross-device tracking language.
- Add a privacy compliance checkpoint to CI pipelines, content workflows, or Openclaw Skills agent routines.
- Support China-based teams launching children’s products in the US market.
How COPPA Guard Works
- The agent or developer provides text through a command-line argument, standard input, or an automated pipeline.
scripts/guard.pyloads the explicit terminology rules fromscripts/rules/terms.py.- The scanner performs local, deterministic text matching across six COPPA-related categories.
- Each match is assigned a category, severity, evidence term, remediation suggestion, and text position.
- Overlapping findings are deduplicated while preserving the higher severity and longer match.
- The tool calculates the overall decision and highest detected risk level.
- Results are returned as human-readable output or structured JSON for Agent and Openclaw Skills consumption.
- Teams use high-risk findings to block or revise publication, while medium- and low-risk findings prompt human review.
COPPA Guard Setup
COPPA Guard has no third-party package requirements. Use Python 3 and run the script from the skill directory.
# Verify Python is available
python3 --version
# Scan text directly
python3 scripts/guard.py --text "Our kids app collects children's data without parental consent"
# Read text from standard input for pipelines or Agent calls
echo "preschool game for toddlers" | python3 scripts/guard.py --stdin
# Return structured JSON for automation
python3 scripts/guard.py --text "Your text here" --format json
# List supported detection categories and evidence
python3 scripts/guard.py --list-categories
For Openclaw Skills integration, invoke the command before publication, pass the draft text as input, and configure the workflow to stop or request review when risk_level is high. No API keys, network access, MCP server, dynamic execution, or additional installation steps are required.
COPPA Guard Data Schema & Taxonomy
The skill is organized around a small, auditable file layout:
| Path or field | Purpose |
|---|---|
scripts/guard.py |
Local detection engine and command-line interface |
scripts/rules/terms.py |
Explicit terminology rules, categories, severities, evidence, and remediation guidance |
requirements.txt |
Empty; the scanner uses only Python standard-library components |
decision |
clean when no findings exist, or flagged when matches are detected |
risk_level |
Highest result severity: high, medium, low, or none |
finding_count |
Number of deduplicated findings |
findings[] |
Individual matches with terminology, category, severity, recommendation, and position |
The v1.0.0 taxonomy contains six detection categories:
- Child-directed content: Signals that a product is intended for children, such as kids app, preschool, or children’s game; typically medium risk.
- Collection of children’s personal information: Data collection or under-13 account language; high risk.
- Behavioral advertising to children: Personalized or targeted advertising language; high risk.
- Missing verifiable parental consent: Statements indicating no parental consent is required; high risk.
- Third-party disclosure: Sharing or selling children’s data; medium risk.
- Persistent identifier tracking: Tracking children across sites or devices; medium risk.
Input text is processed in memory and is not persisted, uploaded, or used for telemetry. The scanner does not generate external files unless the caller redirects its output.
COPPA Guard Advanced Features
- Structured JSON output enables deterministic Agent, CI, batch-scan, and Openclaw Skills integrations.
- Standard-input support makes the scanner suitable for content pipelines and automated publication gates.
- Intelligent overlap deduplication retains the higher severity and longer matching term to reduce noisy reports.
- Rule-engine separation allows maintainers to add or revise terms in
scripts/rules/terms.pywithout changing detection logic. - Explicit, readable rule definitions support security review, compliance review, and reproducible results.
- Fully local execution provides zero network requests, zero telemetry, no background process, and no persistent storage.
- Minimal-permission design uses no external APIs, MCP dependencies, dynamic imports,
eval, orexec. - Human-in-the-loop escalation lets teams block high-risk releases while routing ambiguous medium-risk findings for confirmation.
- MIT-licensed, plain Python implementation is easy to inspect, fork, and embed into internal Openclaw Skills tooling.
- The scanner is a pre-release checklist, not legal advice, a complete COPPA audit, or proof of compliance; final decisions should involve qualified counsel where appropriate.
SKILL.md
Loading
Related Openclaw Skills
QuickClass Quiz Converter
QuickClass Quiz Converter transforms Word documents, PDFs, scanned worksheets, and images into structured QuickClass homework and quiz JSON.
Account Video Downloader
A command-line skill that retrieves, analyzes, and batch-downloads public account content from Douyin, Kuaishou, Bilibili, and YouTube.
WorkBuddy Data Directory Migration
Relocate the complete WorkBuddy data directory to a non-system drive with a Windows junction while preserving existing paths and runtime compatibility.
1F3EA Marketplace
1F3EA Marketplace gives AI agents a controlled way to browse, trade, and participate in an agent-only digital marketplace.
AI Tech Insight Compass
AI Tech Insight Compass turns daily AI trend research into source-backed, publication-ready technical articles.
arXiv Decision Intelligence Report
An evidence-first Openclaw Skills workflow that transforms arXiv literature into validated field intelligence, technical options, maturity assessments, and actionable decisions.
Featured*
