feat(mcp): rank capability search lexically

[reachjalil]

Summary

• rank capability search results with a dependency-free lexical scorer over the full native catalog
• normalize camelCase, conservative stems, stopwords, synonyms, rarity, coverage, and adjacency
• rerank current REST, admin, external MCP, marketplace, and skill matches without changing their public result contracts
• preserve request and argument schemas, invocation metadata, connection actions, and marketplace requirements

Why this version

The useful idea from #2573 is preserved, but the implementation is rebuilt on the current capability-search pipeline instead of reviving stale snapshots. Legacy source scores are only a deterministic tie-breaker, and actionable connection status remains ahead of ordinary matches.

Checks

• bun test test/google-workspace-capabilities.test.ts test/mcp-capability-ranking.test.ts test/mcp-capability-search.test.ts test/mcp-agent-timeouts.test.ts — 49 passed
• pnpm --filter @openwork-ee/den-api exec tsc --noEmit -p tsconfig.json
• git diff --check

Risk and gaps

• lexical relevance is intentionally heuristic and covered by pinned fixtures for high-value queries
• backend-only change; no UI manual journey applies
• full Den API suite was not run in this branch

Supersedes #2573.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
openwork-landing	Ready	Preview, Comment, Open in v0	Jul 21, 2026 2:02pm

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
openwork-diagnostics	Skipped		Jul 21, 2026 2:02pm

[vercel]

@reachjalil is attempting to deploy a commit to the Different AI Team on Vercel.

A member of the Team first needs to authorize it.