API / MCP

total_changesets, total_objects, and avg_objects for a date range. Defaults to the last 7 days if no dates are given. Same as the REST API's GET /api/changesets/summary/

Time series of changeset volume, optionally split into up to 20 named series via group_by. Bucket width is auto-picked to target ~300 points for the given range (hourly for ranges up to ~12.5 days, daily beyond that) — override with interval=hour|day. interval=hour on a range wide enough to blow past the point budget is rejected (400) rather than silently truncated. The grain actually used is always reported back in the response's top-level interval field. Defaults to the last 7 days if no dates are given. group_by=action|object_type|feature splits objects changed by what the replication diffs say about each object (requires metric=objects; objects count on their changeset's day, so the series add up to the metric=objects total); those diffs only go back a few months, and the response's `objects_since` gives the first day covered (earlier days count nothing). With metric=objects, action/object_type/feature count only the matching objects (also from the diffs), e.g. group_by=contributor&action=modify: who modified the most. Same as the REST API's GET /api/changesets/timeseries/

Top N (default 20) contributors/editors/imageries/locales/countries/editor-versions/hashtags, ranked by changeset count or by objects changed, for a date range. Defaults to the last 7 days if no dates are given. dimension=editor_version requires an editor filter (created_by is unbounded across all editors, so it is only ever grouped within one already-selected editor family). dimension=action|object_type|feature ranks objects changed by what the replication diffs say about each object (requires metric=objects; the response's `objects_since` gives the first day those diffs cover). With metric=objects, action/object_type/feature count only the matching objects, e.g. dimension=contributor&action=modify (not with editor_version or hashtag). Same as the REST API's GET /api/changesets/toplist/

Histogram of changeset sizes (objects changed per changeset) in roughly logarithmic buckets, with changeset and object totals per bucket; exact p50/p90/p99/max sizes; and `top_1pct_objects_share`, the share of all objects changed by the largest 1% of changesets. Percentile = smallest size whose cumulative changeset count reaches that fraction. Defaults to the last 7 days if no dates are given. Same as the REST API's GET /api/changesets/distribution/

Exact p25/p50/p75/p90/p95 changeset sizes (objects changed per changeset), plus changeset and object totals, per group. by=contributor|editor|imagery|language|country returns the top N names by changeset count (untagged changesets left out); by=day returns every day of the range, in order. Defaults to the last 7 days if no dates are given. Same as the REST API's GET /api/changesets/distribution/breakdown/

The N largest changesets of the range, by=objects (objects changed, default) or by=area (bounding box area in km², on a spherical Earth; changesets without a bbox left out). Ties are broken newest first. Defaults to the last 7 days if no dates are given. With by=objects, action/object_type/feature rank changesets by their matching objects only (from the replication diffs, from `objects_since`), given in each result's `objects` next to its whole `changes_count`. Same as the REST API's GET /api/changesets/largest/

Changeset count and objects-changed, bucketed into geohash-derived grid cells (see `lat_size_degrees`/`lon_size_degrees` in the response; each cell's `cell` is its geohash, usable with /api/changesets/geo/cell/), for a date range. Defaults to the last 7 days if no dates are given. resolution=fine requires `bbox` (the viewport to scope cells to) and returns a finer grid than the default resolution=coarse. Changesets with an unreliably large bounding box (see the view docstring) are excluded from `cells`. Same as the REST API's GET /api/changesets/geo/

Changesets of one /api/changesets/geo/ cell (a changeset belongs to the cell containing its bounding-box center, like the map counts), newest first, paginated with limit/offset; `has_more` says whether another page exists. Same filters and date range as /geo/, so the cell's `count` there is this list's total. Defaults to the last 7 days if no dates are given. Same as the REST API's GET /api/changesets/geo/cell/

The objects with the most edits over the last 7 days up to now (a fixed window: start_date/end_date are ignored), ranked by edits, ties by type and id. An edit is an object version after the first (a modify or a delete), from the replication diffs. The dimension filters apply to each edit's changeset. Per object: edits, distinct contributors and changesets, the last edit, the latest version, and its latest name tag ('' if none). Same as the REST API's GET /api/objects/most-edited/

Paginated list of ingested changeset records, newest first, with optional filters. Always bounded by a date range — defaults to the last 24 hours if start_date/end_date are omitted (there's no unfiltered "everything" mode; use /api/changesets/timeseries/, /summary/, or /toplist/ for aggregate questions over a large range). Same as the REST API's GET /api/changesets/

Up to 10 distinct known values for a filter field, matching a partial query — backs the dashboard's filter inputs. Same as the REST API's GET /api/autocomplete/