NCERT-grounded content platform: search, explain, generate questions, question papers, and curriculum graphs for Indian CBSE textbooks
Strong tool naming (all verb-prefixed) and comprehensive descriptions (avg 180 chars, well above 34-char baseline). All 13 tools have clear, actionable descriptions explaining WHAT, WHEN, and dependencies. Input schemas present with types and descriptions for all parameters. However, output schemas are not explicitly documented in the source code, responses are inferred from implementation. Parameter descriptions are thorough but lack explicit format constraints (e.g., no regex patterns, min/max for numeric ranges). Error handling guidance is minimal; tools do not document recovery paths or categorize errors. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) visible in schema definitions.
Generate an NCERT-grounded explanation for a topic using retrieval-augmented generation. Internally searches the database for relevant chunks, then produces a structured Markdown explanation with sections: Definition, Key Concepts, Examples, Summary, and optional Mermaid diagrams. The tone adapts to the grade level (simple for grades 1–5, formal for 9–10, analytical for 11–12). Returns the explanation text plus metadata about source chunks and model used.
Generate a single CBSE-pattern question grounded in NCERT content. Internally retrieves relevant chunks, then produces a question with optional answer key and explanation. Supports multiple question types (MCQ, short-answer, long-answer, numerical, diagram-based). Bloom's level and difficulty are inferred from the topic's curriculum metadata unless overridden.
Generate a complete CBSE exam question paper for a specific grade, subject, and exam template. Internally retrieves relevant content chunks, then generates questions across all sections (MCQ, short-answer, long-answer, etc.) matching the exam format. Returns the full paper as Markdown with sections, questions, marks distribution, and optional answer key.
Return the full extracted plain-text of a single NCERT chapter, reconstructed from database chunks. Use list_topics first to find valid chapter numbers for a given grade + subject. The response includes the full text, grade, subject, chapter number, and source type.
Output schemas not explicitly documented. Tool descriptions state what is returned (e.g., 'returns an array of objects with chapter number, chapter_title, and book_title') but no formal JSON Schema definitions are visible in the source code. LLMs cannot reliably parse unstructured responses or plan downstream tool calls without documented output types.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) in schema definitions. All tools are read-only (safe to retry), but this is not declared in the tool metadata. Agents cannot distinguish safe tools from destructive ones without explicit hints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 64 | 2026-07-28+ | v2 |
Return source metadata for a specific chapter: the NCERT download URL, book_code, and local filename. This is a fast lookup with no content parsing. Useful for citations and source attribution.
Return the full curriculum map for a textbook: every chapter's topics with their Bloom's taxonomy levels, difficulty ratings, and chunk counts. Useful for understanding available content before generating questions or papers. Requires exact grade and subject — call list_books first if unsure.
Return a recommended learning sequence from a starting topic to an ending topic. Uses curriculum graph traversal to find the shortest or most pedagogically sound path. Includes intermediate topics, their Bloom's levels, and suggested resources.
Return the prerequisite topics for a given topic. Uses a curriculum graph to identify which topics must be learned first. Useful for scaffolding learning paths and understanding dependencies.
Return every NCERT textbook in the database. Use this FIRST to discover which grade + subject combinations exist before calling any other tool. Each result includes grade, subject, book_code, book_title, and chapters_in_database. The 'subject' values returned here (e.g. 'Mathematics', 'Science', 'English', 'Social_Science', 'Physics_1') are the EXACT strings you must use in all other tools — do not rephrase or abbreviate them. Grades range from 1 to 12.
Return all supported exam paper formats with their total marks, duration, and section breakdown (question types, count per section, marks per question). Call this before generate_question_paper to show the user available options. No parameters required.
Return chapter numbers and titles for a specific textbook. Requires exact grade and subject. Call list_books first if unsure which subjects exist for a grade. Returns an array of objects with chapter number, chapter_title, and book_title.
Full-text keyword search (SQLite FTS5) across all NCERT chapter content. Returns matching chapters ranked by BM25 relevance with a 300-character text snippet. IMPORTANT: grade and subject are strict SQL filters — if provided incorrectly, results will be empty. Omit them to search across ALL grades and subjects, then narrow down after seeing results.
Full-text search (SQLite FTS5) over fine-grained NCERT content chunks. Each result includes the matching text, topic label, Bloom's taxonomy level, difficulty rating, and pre-filled parameters you can pass directly to generate_explanation, generate_question, or get_learning_path. IMPORTANT: grade, subject, and bloom_level are all strict SQL filters — if any value is wrong, you will get zero results. Omit optional filters when uncertain.
Numeric parameters lack explicit min/max constraints. 'grade' accepts 1 - 12 but no minValue/maxValue in schema. 'top_k' defaults to 5 or 8 but no maximum specified. LLMs may pass invalid values (grade=0, top_k=10000) without validation guidance.
Error handling and recovery guidance absent. Tools do not document what happens on invalid input (e.g., non-existent grade, malformed subject string). No guidance on retryability, user-fixable errors, or fallback strategies.
Parameter dependencies not explicitly documented. 'subject' parameter in multiple tools states 'Exact subject string from list_books' but does not warn that passing an incorrect subject will silently return empty results (as noted in search_content description). Agents may not realize they need to call list_books first.