---
summary: "worldbank_get_data reports archived/retired indicators by name instead of blaming valid country codes, and accepts comma- or semicolon-separated countries strings like the other tools."
breaking: false
security: false
---

# 0.3.3 — 2026-09-12

## Changed

- **`worldbank_get_data`'s `countries` string now splits on commas and semicolons**, matching `worldbank_get_poverty` and `worldbank_search_projects` (#25). `"US,JP,KR"` resolves to `["US","JP","KR"]`; a value that splits to zero codes (`","`, `[","]`) is rejected the same way an empty value already was. `"all"` combined with any other code is now rejected before the request, since upstream answers that mix with the same envelope as an invalid code.

## Fixed

- **An indicator the data endpoint won't serve no longer blames the country codes** (#24). The World Bank data endpoint rejects some catalogued indicators — every WDI Database Archives ID, plus non-archive sources like PEFA and ICP — with message id `175`, distinct from the generic id `120` used for a genuinely invalid indicator or country. `worldbank_get_data` now reports these as `indicator_not_queryable`, naming the indicator's catalog source(s) when the lookup succeeds, with recovery pointing at `worldbank_search_indicators` instead of `worldbank_list_countries`.

## Dependencies

- `@cyanheads/mcp-ts-core` ^0.12.3 → ^0.12.8
- `zod` ^4.4.3 → ^4.6.0
- `@biomejs/biome` ^2.5.10 → ^2.5.12 (dev)
- `ignore` ^7.0.6 → ^7.0.9 (dev)
- `tsc-alias` ^1.9.2 → ^1.9.4 (dev)
- `vitest` ^4.1.11 → ^5.0.0 (dev)
- `engines.bun` >=1.3.0 → >=1.4.0