Data catalogue
Find datasets, series, publications and extracts across the marketplace.
- Warehouse rows
- 6.17B
- Structured DB
- 1,711.35 GB
- Live data store
- 189.73 GB
- Data tables
- 1,03,091
- Publications & data products
- 1,749
About these figures
BigQuery reports 6,167,191,751 physical rows across 89 native base tables in 9 catalogue warehouses, measured on . Structured DB measures logical bytes across the same tables. The headline excludes 22 verified staging and backup tables with 365,935,268 rows and 57.49 GB of logical storage. Separate publisher editions can contain overlapping observations.
The earlier 40-MCP inventory reported 1,115,017,091 records across warehouses, files and upstream sources, including 811,051,329 upstream survey records from MoSPI NADA.
Live data store measures 189.73 GB of live catalogue objects across 39 Cloud Storage buckets, measured on .
UNCTAD ingestion files account for 153.97 GB of additional live storage. Retained soft-deleted objects account for 109.84 GB. Raw publisher archives and their corresponding BigQuery tables are measured in their respective storage layers.
Data tables counts 1,03,091 publisher tables held across the marketplace MCPs, measured in each MCP's own store on 2026-11-10: one table, sheet, dataflow, report section or API dataset in one edition counts once, the way data.gov.in counts resources. Raw copies, backups and archive duplicates are not counted twice. They belong to 1,749 publications and data products — six Economic Survey reports are one publication — which hold 10,664 editions, 8,62,701 series and 19,717 documents. India Intelligence and Survey chapter packs are built from these MCPs and are not added. Geo-Spatial can also reach 2,36,742 data.gov.in resources live; they are not held and not counted.
The search index includes 18,933 datasets (tables, survey studies and map layers) and 493 individually indexed series. Separate live and archived access routes can have their own entries. These figures apply across the catalogue, regardless of filters.
Reference & capabilities
74 results
list values
The value ledger — every candidate an agent recorded (status: accepted | parked | all), filterable by entity id, metric id and period key.
make chart
Vega-Lite chart of a finding's series (window shaded) or any selector.
outline article
Answer-first outline: direct answer, then one section per sub-question whose points cite fact, evidence and hypothesis ids.
plan research
Competing hypotheses per kept finding (with expected signatures in the data), then scoped research briefs: one per finding plus context briefs for sub-questions that need outside research. Each brief lists required capabilities and the registered sources they resolved to. `sources` restricts this session to those registered source names (see list_sources), e.g. ["claude-web", "tcs-mcp"]; omit to u
plan retrieval
Batch the cell grid into retrieval tasks (entity × metric domain × period run) and route each to the registered sources whose capabilities fit (SQL and structured sources first). `sources` restricts this dataset to those source names.
preflight dataset
Check LLM readiness and source routing without spending retrieval calls. Reports host-mode alternatives. Use discover_sources to probe connectivity.
preflight insight
Inspect prerequisites, pending research, quality and the next tool to call. Read-only and free of LLM/network calls. Does not launch jobs or follow-ups.
preview table
Quick look at filled values: entity × last N periods for one metric (or each).
profile dataset
Per table: time axis (frequency, range), measures with units and ranges, dimensions with their levels, and each column's inferred role.
query data
Read-only DuckDB SQL over the session's tables. Normalised tables carry _period_start/_period_end/_period_label when a time axis was found; raw copies are raw__<table>.
rank explanations
Test every hypothesis for a finding (or all) and rank them; reports the share of the move quantitatively explained and the unexplained residual.
reconcile
Fill each cell from its best candidate (source tier, then date, then locator precision), keep disagreeing candidates as conflicts, compute derived metrics.
register source
Plug an MCP in as a data source — e.g. a SQL query MCP (capability sql_query), a statistics API (structured_data), a web search MCP. HTTP (`url`, `headers`) or stdio (`command`, `args`); ${ENV} references allowed in headers/env; capabilities inferred from its tools if omitted.
register source
Plug an MCP server in as a research source — HTTP (`url`, optional `headers`) or stdio (`command` + `args`). Header/env values may reference ${ENV_VAR}s. Omit `capabilities` to infer them from the server's tools. `rung` orders sources within a capability (lower = tried first; put browsers last). `scope_entities` limits the source to research about those entities.
remove source
Unregister a source (the built-in claude-web source is disabled instead).
remove source
Unregister a source (the built-in claude-web source is disabled instead).
render article
Final Markdown and HTML with footnotes, charts and a method/evidence appendix.
research coverage
Status of every sub-question (answered / partial / gap). Gaps get one narrower follow-up brief per research round until max_rounds (1-6; default 2 = one follow-up); after that they are disclosed as open rather than widened. With return_pending (a host that runs the briefs itself, and may ask again), a brief planned for the sub-question but not yet run is returned again rather than a follow-up plan
reset insight research
Archive research and writing before editing scope/findings. Retains ingested tables, profile, Charter and analyses. Old evidence is archived, never reused under new hypothesis IDs. Re-select findings and re-plan after reset.
resolve entities
Verify and complete the entity list — official names, aliases, identifiers — against current sources, when the set is defined by a rule ("top 5 by revenue").
Inventory snapshot 15 Nov 2026. Publisher dates and individual source checks can differ.