fix: plainer text for the built-in entries

[datlechin]

Entry

• claude-code, claude-desktop, cursor, zed: description cut to the bridge and the setup step. The capability sentence repeated the Reads and Changes rows.
• Zed: "context server" is now "local MCP server", as Zed's UI says. The context_servers key stays named.
• Same four: disclosures.networkNote is now " sends query results to its model provider" (Claude Desktop: "to Anthropic").
• command-line: drops the inline example commands. Plain text cannot set them as code.
• shortcuts: uses the edition name "TablePro for iPhone and iPad" and splits the last sentence.
• No summary changes, so the web taglines need no update.
• Wording only. No facts, dates or other fields change.

Checklist

• I read the listing policy and the entry follows it.
• python3 scripts/validate.py passes on my machine (also with --online --scan-sources --base origin/main).
• I am the publisher, a member of the publishing organization, or a major contributor.
• The integration uses only documented TablePro interfaces.
• The name does not include TablePro, and the icon is not the TablePro app icon.
• The disclosures are true: what it reads and writes, Safe Mode, network, account, payment.
• I own the icon and screenshots, or may use them, and I let TablePro show them in the directory and its published data.
• I wrote this pull request myself. If an AI tool helped, I say how below and I checked every claim.

Consent or notes

• Six entries in one maintainer PR.
• Open for the owner: the MCP network notes name only query results, while the provider also gets connection names, hosts and schemas from other tools. Cursor may route through its own servers.
• Drafted with an AI agent. Facts checked against the docs on docs.tablepro.app.