FT Test Runner

FT QA agent

UI Test Automation via Claude + Playwright

Total Runs

Pass Rate

Avg Duration

—

Last Run

Pass/Fail Trend

Run History

Saved Batch Reports

No reports yet

No runs match the filter.

Test Cases

Env:

Loading tests...
No tests match filter

Draft Cases

Env:

Agent-proposed QA cases generated from exploration reports. They never sync to TestRail and are excluded from batches / pass-rate. Review, then Run to try one or Supersede to discard. Stale drafts auto-purge after 60 days.

No drafts. Run an Explore and use "Generate drafts" on its report.

No active run. Go to Tests tab to start one.

Steps

Waiting for steps...

No MCP calls yet...

⬇ Download failure tree

Screenshot at failure

Test Cases

Delete Tests

Delete by Section

No sections found

Select all

Select a test case to edit, or click "+ New" to create one.

Created by · Last edited by

Test ID

Portal

Section

Title

Project

Priority

Type

Reference

Preconditions (one per line)

Emulator Serial

Original (ai_step)

Proposed (deterministic)

Generating device parameter tree via bbf-tree-mcp based on test preconditions.

Regenerating emulator_setup from test preconditions and steps using converter logic.

Total

Passed

Failed

Errors

Inconclusive

#	Test ID	Title	Portal	Status	Reason	Time
					—

QA Validation Notes

Behavioural nuances, known backend bugs, and verify-time guidance for QA. Notes saved here are auto-injected into the converter and LLM-verify prompts for tests matching their scope.

include deprecated

Severity	Title	Scope	Status	Updated
No notes match the current filters.

Title *

Body *

/ 20 000 chars

Scope (at least one field required, or set Global)

Test IDs (comma-separated)

Widgets (comma-separated)

KPI types

Topologies

Global — applies to every test

Severity

Status

Tags (comma-separated)

v · created by

Navigation fixes the agent wants to remember

Each row = a navigation self-heal that repeated during runs: the converted test asked for one control, the real UI needed another. Accept stores the fix permanently (portal-scoped, applied at script generation — the test YAML is untouched). Dismiss hides the proposal forever.

Learning endpoints are disabled on the server (FT_LLM_DIAGNOSE is off).

Portal	Converted step → what actually worked	Seen	Cases
No pending navigation proposals — every learned fix is either promoted, dismissed, or no heals have repeated yet.
	→ conflicting not promotable

Recurring failures the agent wants to register as known issues

Each row = the same test failed repeatedly with the same diagnosed durable cause (backend / TestRail / converter bug). Accept adds it to the known-issues registry so future diagnoses classify it instantly. Dismiss hides this signature.

Case	Bucket	Signature / note	Seen
No recurring-failure candidates right now.

Heal journal (raw)

Every runtime navigation heal, grouped. Read-only — proposals above are derived from this.

Case	Step	Target → healed action	Count
Journal is empty — no navigation heals recorded yet.
		→

Validation scan (TestRail)

Environment Defaults

Central environment profiles (an env = a named settings blob). Pick a profile per run via the Env dropdown in the Tests tab; the worker is materialized with that env's settings for the run. Select a profile below, edit the fields, then press Save to store it back centrally on the gateway.

Environment profile

Support Portal URL (e.g. http://ui.test-agent.rd.friendly-tech.com/support-portal)

Management Portal URL (e.g. http://ui.test-agent.rd.friendly-tech.com/management-portal)

Backend UI URL (e.g. http://ui-backend.test-agent.rd.friendly-tech.com)

Used by FtApiClient for monitoring / device-tree calls.

Emulator URL (e.g. http://tr069-emulator.test-agent.rd.friendly-tech.com)

TR-069 emulator HTTP API base (used for direct parameter queries).

ACS URL (e.g. http://ftacs.test-agent.rd.friendly-tech.com/ftacs/ACS)

ACS endpoint the emulated device reports to (Device.ManagementServer.URL). Must point at the same environment as Backend / Emulator URLs.

Configs Service URL (e.g. http://ft-configs-service.test-agent.rd.friendly-tech.com/configs-service)

ft-configs-service — единый писатель настроек приложений на стенде (IspEnable из settings.xml, ispMode из acs_configuration.xml и т.п.). Преконды пишут значения сюда; пусто = на стенде сервиса нет, используется прежний путь через management/ACS REST.

Configs Service user (своя учётка сервиса)

Configs Service password

Username (e.g. admin)

Password (e.g. admin)

ClickHouse URL (e.g. http://rd.friendly-tech.com:30604)

Database (e.g. ftacs_qoe_ui_data)

User (e.g. ftacs)

Password

Type

Host (e.g. logs.example.com)

Port (e.g. 22)

User (e.g. root)

Password

Log directory (e.g. /var/log/oneiot)

Error pattern (regex) (e.g. ERROR|FATAL|Exception|Traceback)

After each test, SSH to this server and grep log files (named in "absent from 'server.log'"-style assertions or log_check steps) for errors appended during the run. Warn-only — attaches to the report, never fails a test. Save settings before testing.

NBI base URL (e.g. http://95.216.35.35:8889)

Creator (blank → portal user) (e.g. dedtest)

Creator password (blank → portal pwd)

App ID (optional)

Device-creation tests (FTCreateDevice / addCPE) call PUT /api/Device as a pre-flight to create a pre-provisioned device — with map coordinates (C9616) or without (C9617). Test connection is a read-only reachability check; auth is validated on the actual create. Save settings before testing.

TestRail

TestRail URL (e.g. https://friendlytechnologies.testrail.io)

TestRail Email (e.g. user@friendly-tech.com)

TestRail API Key

Auto-sync Interval (minutes, 0 = off)

User Management

Create User

Username

Password

Role

Username	Role	Created by	Status	Actions
No users found.
			disabled active

Worker Management

Add Worker

Name (optional — auto-assigned if blank)

Name	Env (pinned)	Display	Base URL	noVNC URL	Status	Actions
No workers configured.
					disabled enabled

Activity Log

Time	Username	Role	Action	Target	Path	Status
No activity entries.

Test Procedures

"How to test X on portal Y" documents written by the librarian as drafts. Only approved procedures are served to test execution. Review a draft, then Approve (publish) or Reject (roll back to the last approved version). Editing is not done here — corrections come via a new librarian draft or the REST API.

Name	Portal	Status	Version	Updated
No procedures.

● Connected ● Disconnected