Replace three separate skills (semantic-search, hybrid-search,
vector-database) with a single vector-hybrid-search skill. Mirrors the
consolidation done in the agent-skills-sandbox repo.

The new skill uses SENSE syntax for all API examples (consistent with
Kibana Dev Tools), covers semantic_text + dense_vector paths, hybrid
RRF, reranking, and production optimization.

Made-with: Cursor

…lder

Agent Builder cannot execute client code — only API calls in SENSE syntax
can run via Kibana Dev Tools. This changes the onboarding agent to:

- Output SENSE API snippets (PUT, POST) instead of Python/JS/Go code
- Replace MCP setup with a note that cluster access is already available
- Replace "what language?" with "where will you build the application?"
- Replace ingestion scripts with Kibana UI routing (File Data Visualizer,
  connectors) and POST _bulk for Dev Tools
- Add IDE handoff section for after Elasticsearch resources are set up
- Replace Code Standards with API Snippet Standards

Made-with: Cursor

Adds a use-case-library skill that maps 8 Elasticsearch use cases with
industry examples and technologies. Adapted for Agent Builder:

- Each use case includes an "In Kibana" section explaining what can be
  set up via Dev Tools before moving to the IDE
- Quick Reference table includes a "Set Up in Kibana" column
- Non-Search redirects use Kibana navigation (Management → Spaces)
  instead of abstract doc links
- No LangChain/LlamaIndex references in "What ES does" sections (those
  are IDE-only integrations)
- No routing directives (agent already has the conversation context)

AGENTS.md updated to reference use-case-library for "what can I build?"
questions.

Made-with: Cursor

…ilder-combined

Made-with: Cursor

Made-with: Cursor

- Updated signal table in AGENTS.md: semantic-search, hybrid-search,
  vector-database → vector-hybrid-search
- Updated keyword-search SKILL.md: semantic-search → vector-hybrid-search
- Updated catalog-ecommerce SKILL.md: semantic-search/hybrid-search →
  vector-hybrid-search
- Added json language tags to all SENSE code blocks in vector-hybrid-search
- Fixed MD060 table alignment in signal table (compact format)
- Regenerated all TypeScript from cleaned sources

Made-with: Cursor

The sandbox skill prioritized Jina as the default EIS embedding model
but this was lost when the kibana skills were written from scratch.

- vector-hybrid-search: Jina v3 as default dense model, Jina v5-small
  for cost-sensitive, Jina Reranker v2/v3 for reranking, EIS docs link
- rag-chatbot: EIS verification note with Jina model details

Matches the sandbox's references/vector-hybrid-search and rag-chatbot.

Made-with: Cursor