Vessel Tracking — live worldwide ship positions from AIS.
Two tools with complete input schemas and strong, detailed descriptions that explicitly warn about coverage limitations. Naming follows verb_noun pattern (vessels_in_box, vessels_by_mmsi). Descriptions are 250-350 chars, well above the 10-char minimum and within the 194-char production baseline. Both tools include critical caveats about partial terrestrial coverage and redirect users to alternative tools (chokepoint_status) for dark regions. However, output schemas are not documented in the source, LLMs cannot see what fields to expect in responses. No error handling guidance visible. Parameters are typed and described, but lack enums for window_seconds or validation ranges.
Query live AIS vessel positions by MMSI (Maritime Mobile Service Identity) numbers. Returns vessels heard during a listening window of a few seconds. COVERAGE IS PARTIAL AND TERRESTRIAL — there is no receiver coverage in the Persian Gulf, Strait of Hormuz, Gulf of Oman or Red Sea, so this tool cannot answer questions about those waters (use chokepoint_status for Hormuz/Suez/Bab el-Mandeb transit counts instead). Counts are vessels heard during a listening window of a few seconds, not a complete count of vessels present.
Query live AIS vessel positions in a geographic bounding box. Returns vessels heard during a listening window of a few seconds, not a complete census. COVERAGE IS PARTIAL AND TERRESTRIAL — there is no receiver coverage in the Persian Gulf, Strait of Hormuz, Gulf of Oman or Red Sea, so this tool cannot answer questions about those waters (use chokepoint_status for Hormuz/Suez/Bab el-Mandeb transit counts instead). Counts are vessels heard during a listening window of a few seconds, not a complete count of vessels present.
Output schemas not documented. LLMs cannot see what fields (mmsi, latitude, longitude, speed_knots, etc.) the response contains, forcing them to guess at downstream field names and breaking tool chaining.
No error handling guidance. If a query returns zero vessels, the LLM cannot distinguish between 'no traffic' and 'no coverage' without explicit error classification. The code comments address this internally but tool responses do not expose the distinction.
window_seconds parameter lacks validation constraints. No minimum/maximum specified; LLMs could pass 0, negative, or absurdly large values (e.g. 86400 seconds). Should constrain to e.g. 1-300 seconds.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 52 | 2026-07-28+ | v2 |
Bounding box parameters (south, west, north, east) lack validation. No description of valid latitude/longitude ranges, no check for south < north or west < east. Invalid boxes could cause silent failures or API errors.
MMSI array parameter lacks size constraints. No minimum/maximum array length specified; LLMs could pass 0 MMSIs or 10,000+ MMSIs, causing performance issues or API rejections.