NASA CMR (Common Metadata Repository) MCP. Catalog search across NASA's entire Earth-science data holdings (~10k collections: satellite missions, instruments, datasets) plus the individual data files (granules) inside each collection. Keyless.
Three well-named, read-only tools with clear descriptions (194 - 250 chars, within baseline 34 - 392). All tools have complete input schemas with typed properties and descriptions. However, output schemas are entirely undocumented, LLMs cannot predict response structure or plan downstream calls. Parameter descriptions lack format constraints (e.g., 'YYYY-MM-DD' is stated but not enforced; bounding_box format is described but not validated). No error handling guidance, callTool catches exceptions but returns generic {error: message} without recovery hints. Missing pagination documentation despite limit parameters suggesting result sets. No tool annotations (readOnlyHint, idempotentHint) despite all three being read-only and idempotent.
Get full metadata for one NASA Earth-science collection by its CMR concept ID — summary, DOI/landing page, spatial coverage, temporal range, and access links. Keyless.
Search NASA's Common Metadata Repository for Earth-science dataset collections by keyword, platform (satellite/mission), instrument, time range, or bounding box. ~10k collections covering MODIS, Landsat, Sentinel, VIIRS, GPM, and every NASA DAAC. Sorted by usage. Keyless.
List the individual data files (granules) of a NASA collection — newest first, with timestamps, file size, and direct download URLs. Filter by time range or bounding box. Keyless.
Output schemas completely undocumented. LLMs cannot predict response structure (fields, types, nesting) or plan multi-step calls. mapCollection() shows internal transformation but no schema is exposed to the agent.
No error recovery guidance. callTool() catches errors and returns {error: string}, but does not categorize as retryable/user-fixable/fatal or suggest next steps. E.g., 'NASA CMR: 400 Bad Request' gives agent no actionable path.
Parameter format constraints stated in descriptions but not enforced in schema. 'YYYY-MM-DD' format for temporal_start/temporal_end is documented but no JSON Schema format or pattern field validates it. bounding_box format 'minLon,minLat,maxLon,maxLat' is described but not validated.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 61 | 2026-07-28+ | v2 |
Pagination not documented. search_collections and search_granules accept limit (default 10, max 25) but no total_count, next_cursor, or offset returned. LLMs cannot know if results are truncated or how to fetch remaining items.
No tool annotations despite all three tools being read-only and idempotent. Missing readOnlyHint and idempotentHint annotations prevent agents from optimizing retry logic and understanding safety guarantees.