Kapa replaces more of the custom reference: GitHub ingestion, retrieval, and hosted MCP search/read. But GitHub refreshes are hourly, not the reference's push trigger, and Public MCP requires Google/GitHub OAuth. Better search quality remains unproven on this corpus.

Kapa relative to the custom reference

Checked September 29, 2026. This maps responsibilities; it is not a benchmark. See the earlier Kapa assessment.

The GitHub connector accepts Markdown and MDX and allows source selection. The refresh schedule lists hourly checks for additions, updates, and deletions. This replaces an owned Git ingestion job with managed synchronization, but not the same event-triggered update contract.

The hosted MCP supplies search and an optional full-document tool. Public access requires Google or GitHub OAuth. Public and internal configurations use default retrieval, not the API-key deep mode.

Relative to the reference, Kapa takes over the Git connector, managed retrieval, and MCP serving. Configuration, corpus selection, evaluation, and commercial/access-policy decisions remain with the Operator.

My interpretation: its main documented distinction is outsourcing more application responsibilities. Whether that produces better retrieval, lower total cost, or acceptable freshness requires evaluation against the same corpus and questions.