Update jev-mcp (jkudish) entry for the 0.10.0 toolset and shipped skill

[kittimzhe]

What

One-line update to the existing jev-mcp (jkudish) entry in categories/infra-sdks-integrations.md — no new entry, no cross-posting.

Why

The entry still describes the project as a "proof-of-concept MCP server" with three capabilities (claim verification, content screening, candidate ranking). The project shipped 0.10.0 today and the description has drifted from reality:

• eleven judgment tools (jev_verify, jev_screen, jev_noul, jev_find, jev_rerank, jev_classify, jev_decide, jev_compare, jev_extract, jev_review, jev_gate), all with fail-closed handling (unknown arguments rejected, malformed provider answers surface as invalid_response rather than a clean pass)
• an agent skill shipped inside the npm package (since 0.9.0), so coding agents get judgment-tool policy — which tool to call when, fail-closed handling, privacy — without hand-written prompts; likely the most list-relevant development
• opt-in stateless Streamable HTTP transport (0.10.0)

The new sentence keeps one line, names the decision layer (eleven tools, fail-closed), and calls out the shipped skill.

Format compliance

• Category file only; the README aggregate is regenerated by scripts/build-readme.py
• Single bullet line, no star badges, no counts
• The Noul tool name matches the vendor's question-type naming

Disclosure

Prepared with the help of an AI coding agent (per CONTRIBUTING, "Disclose AI generation"). I'm a merged contributor to jev-mcp (#36, shipped in 0.10.0), which is how I noticed the entry had drifted; this PR updates the upstream project's entry, not one of my own.