Endpoint Editor Panel

The endpoint editor is the right-hand panel of the connector editor's home view. It shows one endpoint at a time, with an editable header, a vertical tab bar, and a content area. This page documents every part of it.

When you switch to a different endpoint, the panel resets to the Response tab and clears any open popout.

Header bar

Across the top of the panel:

• Method badge (colour-coded: GET / POST / PUT / PATCH / DELETE).
• Title — the endpoint's friendly name. Click the text or the pencil to edit inline.
• Description — a second line under the title; click to edit ("Add a description…" when empty).
• Save — appears while you are editing the title or description; commits both. Pressing Enter in either field also saves; Esc cancels.

Vertical tab bar

A narrow rail on the left of the panel. Tabs, top to bottom:

Tab	What it shows
Response	The test controls and the response/variables split (the default tab).
Config	Endpoint configuration, request body, conditions, transformations, and query/path parameters.
Paging	Pagination and recursion configuration.
Headers	Per-endpoint request headers.
Curl	A copyable cURL command for the current request.
AI Guide	AI discovery findings, the Examine tool, and the vendor docs link. Marked with a sparkle icon.

Non-obvious tab states:

• The Paging tab turns red when pagination is enabled but its configuration is incomplete (and you are on another tab) — a prompt to go finish it.
• The AI Guide tab label is tinted purple.

When the Response tab is active, two extra buttons appear pinned at the bottom of the rail:

• History — toggles the response history popout (see below).
• Edit Raw — placeholder; the raw-editing functionality is not yet implemented.

Response tab

The main testing surface. From top to bottom:

URL bar and test controls

• URL bar — a read-only field showing the platform host and the endpoint path. Hovering shows the fully resolved URL (with {{variables}} substituted using static values and runtime test values).
• Status icon — reflects the last test result (success / error / loading).
• Test button (play icon) — runs the endpoint once. Disabled until test credentials are set; in that state a Set test credentials button appears alongside, linking to the Test tab.
• Full collection toggle — appears only when pagination or recursion is enabled. When ON, the Test button follows all pages/recursion; when OFF, it fetches a single page (capped at 3 pages). The tooltip states the current mode.
• Iterate button (layers icon) — appears only when the endpoint has an iterate-mode dependency mapping. Runs the endpoint once per unique source value via a streaming test; while running it shows live progress (current/total) and clicking it re-opens the progress modal. The tooltip shows how many unique (and total) values will be used.

Parameters and body

• Parameters panel — path and query parameters that feed the URL; hidden when the endpoint has none. Values can come from manual entry, dependency mappings, or variables.
• Body (JSON) — shown only for POST/PUT/PATCH. A JSON editor supporting {variable} placeholders; any {param} tokens become linkable drop zones for dependency mapping.

Result metadata and banners

After a test, a metadata line shows response time, the status code (green for 2xx, red otherwise), page/item counts (when paginated), and the request URL (hover for the full value). Context banners may appear:

Banner	Meaning
"Showing example from API spec" (blue)	No live test yet; the viewer is showing an example pulled from the API spec.
"Using sample data for dependency mapping" (amber)	The displayed response is saved sample data, not a fresh live call.
"Showing error response (sample data preserved)" (red)	The last call errored; your previous good sample is preserved underneath.
"Recursive traversal: …" (purple)	Summary of a recursive run (total requests, max depth, nodes collected).

An auth failure (401/403) surfaces a Check test credentials shortcut to the Test tab.

Response / Variables split

Below the controls, a horizontally split, draggable layout:

• Response (left) — the interactive JSON viewer. Expand/collapse nodes, click a path to select it for a dependency, drag a value onto a parameter, and save selections as variables. You can also paste/set sample data here.
• Variables (right) — the variables panel for this connector (extracted, merged, paired, static, and runtime variables).

Config tab

A two-column layout:

• Left column:
  • Endpoint config — core fields and a per-endpoint rate limit override.
  • Body parameters — for POST/PUT/PATCH: a content-type selector (None / JSON / Form Data / URL Encoded / Raw) and a body editor. JSON bodies expose {param} drop zones.
  • Conditions — execution conditions (run the endpoint only when conditions match), with all/any logic. See Endpoint Conditions.
  • Transformations — the response-transformation chain editor (shown when transformations exist). See Response Transformations.
• Right column: the parameters editor — define path and query parameters for the endpoint.

Paging tab

Hosts the pagination configuration panel and the recursion configuration. When pagination is toggled off, the form body is greyed out and blocked, but the panel's own enable toggle stays clickable so you can turn it back on. See Pagination Reference.

Headers tab

A per-endpoint header editor. These headers are sent in addition to the connector's default headers. It also shows the headers actually sent on the last test, and (if wired) can save headers back to the connector.

Curl tab

Renders a copyable cURL command reflecting the endpoint method, the resolved URL, and the most recent request/response, so you can reproduce the call outside Traces.

AI Guide tab

Help and AI research for the endpoint:

• Examine for variable candidates — runs an AI pass (Claude Sonnet) over this endpoint's response, the execution order, and other endpoints' parameters, then suggests which response fields could feed another endpoint as a variable. It is research only — you still create the mappings yourself.
  • The Examine endpoint button is disabled until you have run a test (there must be a response to examine). Once examined, it shows the suggestion count with View suggestions and Re-examine buttons. Re-examining asks for confirmation first.
  • The Examine modal streams progress, then lists recommendations; clicking a recommendation can focus the relevant endpoint. The modal cannot be closed while a run is in progress (so the stream is not lost).
• Vendor documentation — a link to the connector's documentation URL (or a note if none is set).
• Category / flags — chips for the endpoint's category, Forensically relevant, and Modifies data (when set).
• Reasoning and AI verdict — the LLM's recorded reasoning and a Recommended / Low-priority verdict, when the endpoint was AI-generated. If there are no AI findings, an empty state is shown.

Test result modals

Longer-running tests open dedicated modals:

• Full collection modal — opens when you run a full-collection test; shows live progress and a final summary (status, time, request URL, pagination result, sent headers/body).
• Iterate / unified streaming modal — for iterate runs; streams per-value progress with pause / resume / stop controls. Stopping preserves the partial data already collected.

See Test Modes and Testing Connectors for how to use these.

History popout

The History button (Response tab, bottom of the rail) opens a popout listing this endpoint's previous runs (newest last), labelled by run type (single / full collection / iterate) with timestamps. Selecting an entry loads that past response into the viewer. This is the per-endpoint run history; for saved sample responses see Sample History.