Privacy Policy — Reecopedia MCP Connector

Effective date: 22 August 2026 (previous version: 31 July 2026) · Applies to the public MCP connector at https://ia.reeco.eco/mcp

Read-only search · Optional feedback report · No account · No login · No personal data required

In one line: the Reecopedia connector receives the text of your search query — plus the name and version your AI client announces when it connects (which identifies the software, not you) — and, only if you choose to report a wrong answer, the text of that report. It does not ask for, receive, or store any personal information. Query text is retained without any personal identifier, solely to improve search quality, for at most 90 days.

1. What this connector is

Reecopedia is a public regulatory search engine for EU textile sustainability law — ESPR (Ecodesign for Sustainable Products Regulation 2024/1781), the Digital Product Passport, CSRD, CBAM, EU ETS, CWA 18291 (CIRPASS-2), and official JRC preparatory studies. Exposed via the Model Context Protocol (MCP) to any compatible AI client — Claude, ChatGPT, Gemini, or others, it performs retrieval only: it returns the most relevant passages with source-file and page citations. The answer you read is written by your own AI client, using your tokens — no large-language-model calls are made on Reeco's side.

2. Data we collect

Your query text — the search question you (or your AI client on your behalf) send to the search_eu_textile_regulations, search, fetch, document_outline or find_exact_text tools (for fetch and document_outline, the document identifier of a previously returned result; for find_exact_text, the literal string searched).
Your feedback report (optional) — if you (or your AI client at your request) call the report_answer_issue tool to flag a wrong or misleading answer, the connector receives the original query text and the reason you give. Like queries, reports carry no personal identifier; submissions are rate-limited.
Your AI client's name and version — the Model Context Protocol requires every client to announce itself when it opens a connection (for example claude-ai/1.0). This identifies the software you are using, never you: it carries no account, no user id, no session and no IP. The connector has always received it, because the protocol sends it before any search; since 22 August 2026 it is also stored alongside the query text, under the same 90-day retention, so that we can tell which AI clients use the corpus and improve the tools for them. Together with query text — and a feedback report if you send one — this is the only data the connector receives.
No personal data. The connector has no account, no login, and no authentication. It does not receive your name, email, IP-derived identity, conversation history, memory, or files. It cannot request them.
Standard network metadata — like any website, the hosting layer (our EU server and the Cloudflare network in front of it) processes ordinary request metadata (timestamp, that a request occurred) to deliver and protect the service. This transit metadata is not joined to your query text.

3. How we use and store it

Your query text is used to run the search against the indexed corpus and return passages.
The query text is also appended to a retrieval-improvement log that contains no personal identifier — no IP address, no user or session identifier, nothing that could link a query to a person. We use this log to measure retrieval quality and decide which official documents to add to the corpus.
Feedback reports (report_answer_issue) enter Reecopedia's human curation queue: they are reviewed to correct the corpus and improve answer quality.
The log is stored on Reeco's own server located in the European Union.
No profiling, no advertising, no automated decision-making about individuals is performed.

4. Third-party sharing

We do not sell, rent, or share query logs with third parties. The service runs on infrastructure operated by Reeco (an EU virtual private server) with Cloudflare acting as the network/CDN and security layer in front of it. These providers process transit data as processors on our behalf; they do not receive the retrieval-improvement log.

5. Data retention

Query-text logs are retained for at most 90 days, then automatically rotated and deleted. Feedback reports (query + reason, no identifier) are kept until the reported issue has been reviewed and resolved in the curation process. Because logs and reports contain no personal identifier, an individual query or report cannot be traced back to a person.

6. Your rights

We collect no personal data and no identifiers, so there is nothing that identifies you to access, correct, or erase. If you believe a specific query you submitted contains information you would like removed from the corpus-improvement log, contact us (Section 8) with enough detail to locate it and we will delete it.

7. Children

The connector is a professional regulatory-research tool and is not directed to children. It collects no personal data from anyone.

8. Contact

Data controller: Stefano Cipriani — Reeco®, founder, Expert Member of CIRPASS-2 (EWG1, EWG3), JRC Registered Stakeholder.
Email: [email protected] · Web: reeco.eco

9. Changes

If this policy changes, we will update the effective date above and publish the new version at this URL.