Loading ToolBench
Core server for exposing India's open government data (data.gov.in) APIs via the Model Context Protocol, powering agentic AI data exploration.
The server defines 12 data-access tools with explicit schemas and descriptions. Most tools follow verb_noun naming and include parameter type definitions. However, there are consistent gaps: (1) Descriptions vary widely in quality, some are detailed and helpful, others are generic or minimalist; (2) Several tools lack input parameters entirely (beneficiaries_under_pmkisan, district_social_welfare_officers, eight_core_industries, hiv_prevalence with optional params only), limiting their utility; (3) No tool includes explicit output schema documentation in the visible code, LLMs must infer the response structure; (4) Error handling guidance is absent across all tools; (5) No tool annotations (readOnlyHint, destructiveHint, idempotentHint) are declared, even though all are READ_ONLY; (6) Several parameter descriptions reference only field names without explaining the business meaning (e.g., 'Filters the result with field CityCode' does not explain what CityCode represents or why an agent would filter by it). Strengths: consistent naming (verb_noun style), explicit parameter types and ranges (agriculture_growth_rate, daily_market_prices_commodity), thoughtful aggregation and statistical parameters (evapotranspiration, disaster_fund), and pagination support (offset, limit). The schema quality is above-average for community servers, but descriptions lack LLM optimization and output structure is undocumented.
Tool for querying agriculture growth rate data from data.gov.in. Supports filtering by financial year, aggregation by year and growth rate ranges, and statistical analysis.
Tool for querying data on beneficiaries under PM-KISAN (Pradhan Mantri Kisan Samman Nidhi) scheme from data.gov.in.
Tool for querying city-wise wellness centre data from data.gov.in. Supports filtering by various wellness centre attributes and aggregation by city code, category, and city name.
Tool for querying CPI (Consumer Price Index) inflation data from data.gov.in. Supports filtering by financial year, aggregation by year and inflation rate ranges, and statistical analysis.
Tool for querying daily district-wise rainfall data from data.gov.in. Supports filtering by state, year, district, and date range with aggregation and pagination support.
Tool for querying daily market prices of commodities from data.gov.in. Supports filtering by state, district, commodity, and arrival date range with aggregation by district and commodity.
Output schemas are not documented. No tool explicitly defines the structure, fields, and types of the response. LLMs must infer response shape from tool names and descriptions, increasing hallucination risk and making downstream tool-chaining error-prone.
Four tools (beneficiaries_under_pmkisan, district_social_welfare_officers, eight_core_industries, hiv_prevalence) have no or minimal input parameters, severely limiting their filtering capabilities. Tools that do not accept parameters are discovery-only and cannot be used for targeted queries, forcing agents to fetch and filter client-side.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 59 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 41 | - | v1 |
Tool for querying disaster fund allocation data from data.gov.in. Supports filtering by state and aggregation by state, central share, state share, and total allocation with statistical analysis.
Tool for querying data on district social welfare officers from data.gov.in with pagination support.
This provider fetches data from the 'Eight Core Industries' dataset. The API does not support filtering, only pagination.
Tool for querying evapotranspiration and soil moisture data from data.gov.in. Supports filtering by state, district, year, and month with aggregation and statistical analysis.
Tool for querying FCI (Food Corporation of India) commodity stock position data from data.gov.in. Supports filtering by district name, year, month, and commodity name with aggregation by district and commodity.
Tool for querying HIV prevalence data from data.gov.in. Supports optional filtering by state or union territory with pagination support.
No tool includes error handling or recovery guidance. If a tool returns an error (invalid date, missing commodity, state not found), the LLM has no actionable next step. Error messages should include the invalid value, the constraint violated, and suggested alternatives (e.g., 'District "XYZ" not found. Available districts: ABC, DEF, GHI').
Tool annotations (readOnlyHint, destructiveHint, idempotentHint) are not declared. All 12 tools are READ_ONLY and idempotent (safe to retry), but this is not explicitly signaled to the LLM. Explicit annotations help agents reason about safety and retry logic.
Parameter descriptions are often minimal and field-name-focused rather than business-context-focused. Examples: 'Filters the result with field CityCode' (does not explain what CityCode represents), 'The month for which data is to be fetched' (does not explain format: is '1' or '01' or 'January' expected?). Descriptions should explain WHAT the parameter controls, WHEN the LLM should provide it, and what format is expected.
Parameter names use redundant suffixes ('district_name', 'commodity_name', 'wellness_centre_name') when a single parameter ('district', 'commodity', 'wellness_centre') would be clearer. Suffixes like '_name' and '_code' should be reserved for disambiguation only (e.g., user_id vs user_name when both are valid).