API diagnostics for single-page applications

Paxter can capture the API responses that a browser workflow observes. This is useful when a single-page application shows a generic message such as “Unable to save”, but the backend response explains the real problem—for example, that a required field was omitted.

There is no separate SPA browser mode. Enable ApiDiagnostics in the workflow and target the API route that matters. Paxter subscribes to browser network responses before the workflow starts, so it works for fetch and XHR requests made by common SPA frameworks.

Example: turn an API validation error into a regression check

This example fills the two fields available in the UI. The API endpoint is expected to reject the request if a required backend field is missing. A 400 or 500 response causes the Paxter run to fail and preserves the response evidence.

Headless: true
NeedVideo: false
ApiDiagnostics:
  Enabled: true
  # Use the narrowest route that identifies this operation.
  UrlPatterns:
    - "**/api/customer-profiles"
  CaptureResponseBody: true
  FailOnErrorResponse: true
  ResponseWaitMilliseconds: 1000
  MaxResponseBodyCharacters: 8000
  RedactJsonProperties:
    - password
    - token
    - accessToken
    - refreshToken
    - authorization
    - cookie
    - secret
    - clientSecret
    - apiKey
Actions:
  - Do: Page_GotoAsync
    Data1: https://your-approved-test-environment.example/customer-profile
  - Do: GetByLabel_Fill
    Data1: Given name
    Data2: Test
  - Do: GetByLabel_Fill
    Data1: Family name
    Data2: Person
  - Do: GetByRoleButton_ByName_Click
    Data1: Save
  # Wait for the UI result that your users should see. This also gives the API call time to finish.
  - Do: GetByText_Exact_Click
    Data1: Unable to save

For this workflow, a response such as 400 Bad Request with a JSON validation message appears in the HTML report’s API diagnostics tab. The run also becomes a failed regression result because FailOnErrorResponse is enabled.

Network traffic per workflow action

Use NetworkTraffic when you need a lightweight timeline of browser requests, responses, and browser-level request failures for a site. Every event is associated with the Paxter action active when the event was observed. This makes it practical to compare a failed run with a healthy run and see which click or navigation triggered an unexpected route, status, or failed asset/API request.

Headless: true
NeedVideo: false
NetworkTraffic:
  Enabled: true
  # Scope this to an approved site or service. Empty records all browser traffic.
  UrlPatterns:
    - "https://your-approved-test-environment.example/**"
  IncludeQueryString: false
  # A short observation window after every Paxter action.
  ActionWaitMilliseconds: 250
  # Observe work that begins shortly after the final action.
  FinalWaitMilliseconds: 1000
  # Make browser transport failures (for example DNS, CORS, or a disconnected request) regressions.
  FailOnRequestFailure: true
Actions:
  - Do: Page_GotoAsync
    Data1: https://your-approved-test-environment.example/customer-profile
  - Do: GetByRoleButton_ByName_Click
    Data1: Save

NetworkTraffic records metadata only: time, workflow step, action name, event kind (request, response, or requestfailed), HTTP method, route, resource type, status, content type, and browser failure text. It is intentionally not a HAR export and does not store request/response bodies or headers. Use ApiDiagnostics alongside it when you need a bounded, redacted API response body.

Setting	Default	Meaning
Enabled	false	Turns network-timeline collection on.
UrlPatterns	Empty	URL globs to record. * matches within one path segment and ** matches across segments. Empty captures all browser traffic, so scope this in most workflows.
IncludeQueryString	false	Includes query strings in the timeline. Keep it off unless values are safe to retain.
ActionWaitMilliseconds	250	Bounded observation window after each workflow action. Allowed range: 0–10,000 milliseconds.
FinalWaitMilliseconds	1000	Bounded final observation window after the last action. Allowed range: 0–10,000 milliseconds.
FailOnRequestFailure	false	Fails the run when the browser reports a request failure. It does not fail ordinary HTTP 4xx/5xx responses; use ApiDiagnostics.FailOnErrorResponse for that.

When events are captured, Paxter adds a Network traffic tab to index.html, writes network-traffic.json, and includes the entries in output.json. The JSON contains no body or header values and is a useful structured input for regression comparisons or an AI-assisted investigation.

API response settings

Setting	Default	Meaning
Enabled	false	Turns collection on. Nothing is captured unless this is true.
UrlPatterns	Empty	URL globs to capture. * matches within one path segment and ** matches across segments. When empty, Paxter captures all fetch/XHR responses. Prefer one or more narrow routes.
CaptureResponseBody	false	Includes text, JSON, XML, or JavaScript response bodies in evidence. Turn it on only when the response is safe to retain.
IncludeQueryString	false	Includes URL query strings in evidence. Keep this off unless query values are verified safe to retain.
FailOnErrorResponse	false	Fails the workflow if a captured response has an HTTP status of 400 or higher.
ResponseWaitMilliseconds	1000	Final bounded wait for a late response after the last workflow action. Allowed range: 0–10,000 milliseconds. A normal UI assertion after submit is still recommended.
MaxResponseBodyCharacters	8000	Maximum body text written per response. Allowed range: 0–100,000. Truncated bodies are marked in the report and JSON.
RedactJsonProperties	Common secret names	Case-insensitive JSON property names whose values are replaced with [REDACTED] before evidence is saved.

Evidence and AI-assisted investigation

When at least one response is captured, the run folder contains:

• index.html — a readable API diagnostics tab showing method, route, status, response type, and sanitized body.
• api-diagnostics.json — the same sanitized structured response data, intended for a ticket or an AI-assisted investigation.
• output.json — the full Paxter run record, including the API diagnostic entries.

Give an AI the relevant sanitized api-diagnostics.json, the failed UI step, and the intended behaviour. It can then connect a generic UI message with the backend validation result. Paxter records evidence; it does not itself infer the cause or send evidence to an AI service.

Safety and scope

Only use this on systems you are authorised to test. API response bodies can contain personal or confidential data even when common secret field names are redacted. Paxter deliberately does not store request bodies or response headers other than the response content type, but workflow owners must still:

• Use a dedicated test environment and non-production test data.
• Filter to only the endpoint needed for the scenario.
• Keep CaptureResponseBody off unless the diagnostic value justifies the retention risk.
• Add organisation-specific field names to RedactJsonProperties.
• Store the run folder in an approved, access-controlled location and do not commit it.

Do not enable FailOnErrorResponse for an endpoint where a 4xx response is the expected outcome unless that expected response is outside the configured URL patterns.