- Home
- Skills
- Agents & Orchestration
- ai act transparency
ai act transparency
Audit and fix software products for EU AI Act Article 50 transparency, disclosure, and metadata marking requirements.
$29
Works with the AI Agents you already use
ai act transparency
Example session with this skill installed
Our app uses GPT for a support chat and Fal.ai for generating user profile pictures. We are launching in France next week. Audit the code for EU AI Act Article 50 compliance.
- Read your context and instructions
- Compiled the ai act transparency
- Generated the document
I have audited the repository for AI touchpoints related to Article 50 transparency requirements. I found two primary systems requiring disclosure and metadata preservation.
This is a checklist based on Article 50 and the Commission's July 2026 guidelines. It is not legal advice.
Article 50 Audit Findings
| # | Feature | Where | Paragraph | Role | Status |
|---|---|---|---|---|---|
| 1 | Support Chatbot | src/components/Chat.tsx | 50(1) | Provider | HIGH RISK |
| 2 | Profile Gen | src/lib/images.ts | 50(2) | Provider | MEDIUM RISK |
Findings
[HIGH] Support chatbot lacks AI disclosure
- Where:
src/components/Chat.tsx:24 - Finding: The assistant joins the chat and greets the user without disclosing its synthetic nature.
- Fix: Add a persistent disclosure banner and update the initial system greeting.
[MEDIUM] Image metadata stripped in pipeline
- Where:
src/lib/images.ts:12
Finding: The sharp().resize() call does not include .withMetadata(), stripping C2PA/IPTC provenance data from Fal.ai.
- Fix: Update the image processing utility to preserve metadata.
Proposed Code Changes
// src/components/Chat.tsx
// Add AI disclosure notice at the top of the chat window
export const ChatWindow = () => {
return (
<div className="chat-container">
<div className="text-xs text-gray-500 p-2 bg-gray-100">
Vous interagissez avec un assistant IA. (You are interacting with an AI.)
</div>
{/* ... rest of component */}
</div>
);
};
// src/lib/images.ts
// Preserve machine-readable marking for Article 50(2)
export async function processAvatar(buffer: Buffer) {
return await sharp(buffer)
.resize(256, 256)
.withMetadata() // Required to keep provenance markers
.toBuffer();
}
Next steps
- Apply the UI disclosure changes to the
ChatWindowcomponent. - Update the image processing utility to prevent metadata loss.
- Review the
legal-review.mdfile for questions regarding "obviousness" exceptions.
Chat.tsx
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Software teams launching in the EU face fines up to €15 million or 3% of turnover for failing to disclose AI interactions. Identifying every chatbot, generated image, and synthetic voice across a large codebase is prone to human error, and interpreting Article 50 duties for different roles is complex.
This skill audits your repository against the EU AI Act Article 50 transparency requirements. It scans your codebase for model SDKs, media generation APIs, and UI surfaces to build a complete inventory of AI touchpoints. The skill classifies each finding by your legal role—provider or deployer—and checks for mandatory disclosures, machine-readable watermarking, and deepfake labels. It produces a prioritized findings report with specific file references and ready-to-use UI copy for fixes.
What it does
- Inventory scanning identifies AI usage across OpenAI, Anthropic, Gemini, Mistral, and other model providers within your code.
- Role classification determines if you act as a provider or deployer for each feature to assign correct legal duties.
- Gap analysis checks for missing AI disclosures in chat interfaces, voice systems, and public-facing media.
- Pipeline auditing detects where image processing or CDNs might be stripping required C2PA or machine-readable metadata.
- Remediation generation provides code diffs and localized UI wording to bring features into alignment with transparency rules.
How it works
- Trigger the audit by requesting an Article 50 check or providing a target launch market in the EU.
- Review the generated inventory table of AI touchpoints, SDKs, and UI surfaces found in the codebase.
- Analyze the prioritized findings report which ranks risks from high to low with specific file and line citations.
- Apply proposed code changes for disclosure components, metadata-preserving pipelines, and updated UI copy.
Frameworks & tools
Works with major AI SDKs including LangChain, Vercel AI SDK, OpenAI, Anthropic, and Fal.ai. It identifies patterns in React, Next.js, and standard web templates for UI disclosures.
Why this beats prompting it yourself
Generic prompts often miss the technical nuance of how metadata is stripped during image processing or the specific distinction between provider and deployer duties. This skill uses a structured reference set based on the July 2026 Commission guidelines to ensure technical fixes match legal requirements.
Use cases
- Audit a new AI feature for compliance before deploying to users in the European Union.
- Check if existing image processing pipelines are compliant with machine-readable marking deadlines.
- Generate localized disclosure copy for a multi-language chatbot to meet transparency standards.
- Identify prohibited emotion recognition usage in workplace or educational software contexts.
Known limitations
This tool provides a technical checklist and code audit, not legal advice. It cannot identify AI features configured entirely in third-party dashboards without code-level integration.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Checklist only, not legal advice. Cannot detect AI features managed entirely in third-party SaaS dashboards without code integration. Role classification may require manual legal review.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 19 days ago
- Software
Python