The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Global Fishing Watch listing page.
Global Fishing Watch API v3 — AIS-derived vessel identity, at-sea events (apparent fishing, encounters, loitering, port visits, AIS gaps) and aggregated apparent fishing effort.
Part of Pipeworx — an MCP gateway connecting AI agents to 1683+ live data sources.
gfwfish_vessels_search(query, datasets?, limit?, offset?, _apiKey) — vessel
identity by name, MMSI, IMO or call sign.gfwfish_events(start_date, end_date, vessel_id?, event_type?, limit?, offset?, _apiKey) — events over a date range.gfwfish_4wings_report(bbox, start_date, end_date, group_by?, temporal_resolution?, spatial_resolution?, datasets?, _apiKey) — apparent
fishing effort aggregated over a bounding box.Platform-backed since 2026-09-18. PLATFORM_GFW_FISHING_TOKEN is in the
encrypted keystore (GFW API application "pipeworx", non-commercial intended
use); the gateway injects it as _apiKey, and a caller may pass their own
instead. Every v3 endpoint requires a Bearer token — there is no keyless or
demo path. Free non-commercial tokens:
https://globalfishingwatch.org/our-apis/tokens.
Distinct from PLATFORM_GFW_KEY, which is Global Forest Watch — different
organisation, different API, similar abbreviation.
licence: non-commercial is set on the pack: GFW data is free for
non-commercial use and commercial use needs their permission.
A 401 from this API proves nothing about the path. An unauthenticated
request to any path — including one that does not exist, verified against
/v3/zzz-not-a-path on 2026-09-17 — answers 401 {"error":"invalid token"}.
That is why the shapes shipped on 2026-09-17 were unverifiable, and why the
first probe on a real token returned a 422.
The two paging schemes are different, and mixing them is the 422.
/v3/vessels/search pages by CURSOR: the response carries since, you pass it
back as since, and sending offset is rejected 422 Unprocessable Entity.
/v3/events pages the ordinary way — offset and limit are both required,
and the response carries nextOffset. Do not unify them.
Other shape facts, each from the v3 docs and then proved live (fleet #2245):
datasets[0]=...,
vessels[0]=..., filters[0]=..../v3/vessels/search needs at least 3 characters in query; limit maxes at
50. A since cursor carries its own page size — pass since with a
smaller limit and you still get the original page's worth back, so do not
read returned as proof the limit was honoured (measured live: limit 2 plus a
cursor from a limit-5 call returned 5).datasets[0], includes[0]. The bare includes[] form is a 422. Identity is nested under two parallel arrays — selfReportedInfo (what
the vessel broadcast) and registryInfo (what a registry holds) — which can
disagree. The pack lifts the common fields and keeps both arrays./v3/events takes start-date / end-date on GET (hyphenated) but
startDate / endDate in a POST body (camelCase). This pack uses GET./v3/4wings/report is a POST: the grid parameters go in the query string
(spatial-resolution, temporal-resolution, group-by, date-range as
YYYY-MM-DD,YYYY-MM-DD, format=JSON) and the AREA goes in the body — either
a named region, {"region":{"dataset":"public-eez-areas","id":8371}}, or a
custom polygon under a top-level "geojson" key. temporal-resolution is one
of HOURLY, DAILY, MONTHLY, YEARLY, ENTIRE. Its entries are arrays of grouped
rows rather than rows, so the pack flattens them.Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):
tools/list at https://gateway.pipeworx.io/global-fishing-watch/mcp returns the tools in the table
above plus the shared Pipeworx meta-tools — ask_pipeworx,
discover_tools, search_within, remember/recall and the rest of the
gateway-wide set. So the tool count you see is larger than this table: a
single-pack endpoint currently lists roughly 30 shared tools alongside the
pack's own. The connection's initialize response states its exact scope, and
is the authoritative answer for a given day.
This is deliberate, not multiplexing by accident. The meta-tools are what let a
scoped connection answer a question this pack does not cover — via
ask_pipeworx, which routes across the whole catalog — without you adding a
second MCP server. There is currently no way to mount a pack endpoint without
them; if the extra schemas cost you more context than the routing is worth,
connect to the full gateway once rather than to several pack endpoints.
Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:
Both URLs reach the same gateway and the same 1683+ data sources. The
only difference is which pack's tools are listed directly; ask_pipeworx
reaches all of them from either one.
No account needed for the first calls. Inspect any tool: GET https://gateway.pipeworx.io/v1/tools/gfwfish_vessels_search. Find one: POST https://gateway.pipeworx.io/v1/tools/search_packs with {"query":"..."}.
This package also runs as a local stdio MCP server — no Pipeworx account, no gateway round-trip:
Or run it directly to confirm it starts:
It speaks MCP over stdin/stdout and answers initialize/tools/list/tools/call
for only this pack's tools — none of the shared meta-tools the gateway
connection above adds. Same source, same tools, no ask_pipeworx routing.
Instead of calling tools directly, you can ask questions in plain English — this works on the pack endpoint above as well as on the full gateway:
The gateway picks the right tool and fills the arguments automatically.
MIT