Integrate Examozhi APIs

Textbooks are revised on multi-year cycles, and old and new editions often run side by side in classrooms. On Examozhi a subject is a textbook, and multiple editions of the same subject coexist — each with its own chapters, topics, and embeddings. One edition is flagged as current.

• Default behavior: AI RAG search returns the current edition of every subject, so you always serve the live curriculum without doing anything.
• Pin an edition: pass edition (e.g. "2024") on /ai/rag-search or /content/search to retrieve a specific revision — useful for students still on the older book.
• Search everything: set include_all_editions: true (RAG) or current_only: false (text search) to span every edition at once.
• Discover editions: GET /content/subjects?class_id=… returns each subject's edition, is_current, and effective_from so you can offer an edition switcher in your own UI.

# Serve the current edition (default)
POST /ai/rag-search { "query": "laws of motion", "class_id": "…" }

# Pin the 2018 revision for a class still using the old book
POST /ai/rag-search { "query": "laws of motion", "class_id": "…", "edition": "2018" }

Search in one script, retrieve in another. Romanized and code-mixed Indic queries — Tanglish and Hinglish likevanakkam orkabaddi — match native-script curriculum content, on top of the same hybrid (dense + sparse) retrieval that powers /ai/rag-search.

Try it from the Playground — set a target language and enter a romanized query to see cross-lingual retrieval in action.

Some questions are set or ordinal operations, not similarity lookups — “list all the poems”, “show all poem authors”, “summarize the first poem”, “every formula in Chapter 3”. Semantic search answers these with a truncated top-k sample./content/enumerate answers them with the complete, canonically-ordered set (by chapter then position), paginated, with no top-k cap — so an AI agent or your UI gets the whole list, in order, every time.

# "List all poems in the Tamil textbook" — complete, ordered set
curl -X POST https://examozhi.com/api/v1/content/enumerate \
  -H "Content-Type: application/json" \
  -H "X-API-Key: sk_xxxx_your_key" \
  -d '{ "intent_query": "list all poems", "subject_id": "…", "language": "ta" }'

# → {
#     "items": [ { "content_id": "…", "title": "…", "subtype": "poem",
#                  "attributes": { "author": "…" }, "chapter_title": "…",
#                  "page_number": 12, "source_ref": "…" }, … ],
#     "total": 20, "mode": "set", "page": 1, "page_size": 100
#   }

• Modes (returned as mode): set (full list), attribute (distinct values, e.g. authors), ordinal (the Nth / first / last item), chapters, topics.
• No truncation: total is the full match count; page through large sets with page / page_size.
• Versioning aware: honors edition / current_only exactly like search.

The endpoint requires the same Content API entitlement as/content/search (included on every plan with core content access) and is subject to your plan's rate and monthly-quota limits below.

Examozhi ships a native MCP server and a curated agent tool gateway on Pro and Enterprise plans — so AI agents (Claude, Cursor, Pydantic AI, Strands, LangGraph, or your own) can discover and call curriculum tools natively.

# 1) Handshake
curl -X POST https://examozhi.com/api/v1/mcp \
  -H "X-API-Key: sk_xxxx_your_key" -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}'

# 2) Discover tools
curl -X POST https://examozhi.com/api/v1/mcp \
  -H "X-API-Key: sk_xxxx_your_key" -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/list"}'

# 3) Call a tool
curl -X POST https://examozhi.com/api/v1/mcp \
  -H "X-API-Key: sk_xxxx_your_key" -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":3,"method":"tools/call",
       "params":{"name":"rag_answer","arguments":{"query":"Explain photosynthesis"}}}'