Privacy policy
Last updated 30 September 2026.
This policy covers two things built and supported by Michael Costa, trading as White Hat Harbor ("we"): the White Hat Harbor connector for Claude ("the connector"), a remote MCP server at mcp.whitehatharbor.com that we run; and the Echidna Web Data plugin for Claude ("the plugin"), which uses Apify's MCP server. They handle data differently, so each has its own section.
The connector
Data collection
We collect only what the connector needs to know whose daily allowance a request uses:
- Your account id. When you sign in with GitHub, the connector reads your numeric GitHub user id and your username. It requests no GitHub permissions, can't see your repositories or email address, and discards GitHub's access token as soon as it has read the id.
- A sign-in record. The access that Claude holds for the connector is stored with your account id and username, so each request can be matched to your account. The tokens themselves are stored only in hashed form, and the record includes the name and return address that Claude registered for the connection.
- Daily lookup counts. A number per account per day (UTC), for example "7 lookups used on 30 September 2026". It doesn't say which tools you used or what you looked up.
We don't collect your prompts or conversations. The connector only sees the tool inputs Claude sends it (web addresses) and doesn't store them or the results.
Usage
Tool inputs are used only to fetch the pages you asked about and build the answer, which goes back to Claude. The connector fetches public web pages only, from Cloudflare's network, with its own user agent. Before fetching, it looks up the site's address through Cloudflare's DNS resolver (cloudflare-dns.com) to refuse private networks, so the site's hostname is sent there. Job-openings and company-register lookups send the company name or number you asked about to the job board or register being searched (for example Greenhouse, the French register at recherche-entreprises.api.gouv.fr, or GLEIF).
Logs
The connector writes no log lines of its own. It runs on Cloudflare Workers with Workers Logs turned on, so Cloudflare records a standard entry for each request to the server: the time, method, address (on the sign-in pages that includes the one-time sign-in code, which expires within minutes and can't be reused), response status, duration and CPU time, any unexpected server error, and request metadata Cloudflare adds, such as the country it came from. The request body, where tool inputs travel, isn't part of these entries. We use them to find errors, and Cloudflare keeps them for up to 7 days.
Third-party sharing
The connector runs on Cloudflare, which processes requests and stores the data above on our behalf (Cloudflare privacy policy). Sign-in goes through GitHub (GitHub privacy statement). We don't sell, rent or share data with anyone else, and the connector has no analytics, tracking or advertising.
Data retention
Daily lookup counts are deleted automatically two days after the day they count. The sign-in record, with your account id and username, is kept until you ask us to delete it: removing the connector in Claude stops Claude from using it but doesn't delete it on our side. Email us and we'll delete your account's data.
The plugin
The plugin declares one remote connector (Apify's MCP server, limited to our Apify Actors) and a set of skills. It contains no code that runs on your computer, and we operate no servers for it.
Data collection
The plugin collects nothing. We don't receive your prompts, your tool inputs, the results, or your Apify credentials. When Claude calls one of the plugin's tools, the tool input (for example URLs, company names or search terms) goes directly to Apify's MCP server and runs on your own Apify account, which you sign in to through Apify's OAuth page.
Usage and storage
Tool inputs are used only to run the Actor you asked for. Results are stored in your Apify account's storage, under Apify's data retention rules, and returned to Claude. The Actors fetch only public web pages and public APIs, respect robots.txt, and don't log in to any site. As the Actors' developer we can see the aggregate usage statistics Apify provides to all developers; they don't include your inputs or results.
Third-party sharing
Data is shared with Apify only, because the tools run on Apify's platform in your account. Apify's handling of it is described in the Apify privacy policy. We don't sell, rent or share data with anyone else, and the plugin has no analytics, tracking or advertising.
Data retention
We retain nothing, because we never receive any data from your use of the plugin, so there is nothing for us to delete. Apify's retention rules apply to the runs and results in your Apify storage, and you can delete them from your Apify account at any time.
Contact
Questions or requests about privacy: support@whitehatharbor.com.
Changes
We'll update this page and its date if the connector's or the plugin's data handling changes.