Workspace
Explorer, Command Center, a multi-session workbench, Test Paths, run visualization, local issue detection, fix & re-run, an embedded terminal, and Cloud Mode inside TestRelic Studio.
Explorer
Open any local repo — framework auto-detected — and browse its tests in a Postman / VS Code-style tree with detail tabs.
Chat
The primary pane: a persistent Ask AI session backed by the same engines as the CLI (Cloud agent, Claude Code, Codex, Ollama), with tool approvals and an artifacts gallery. Chat is one surface of a larger multi-session workbench — see below for running several sessions at once, scheduled Routines, and one-off Dispatch jobs. See Live Browser for what happens when the agent needs to drive a page.
Found it — the promo field is queried before the cart total re-renders, so the assertion races a stale value. I moved the check onto expect.polland re-ran it 5× clean.
Command Center
The gear icon opens the Command Center — a plain-language settings and command hub, for
testers who've never seen a slash command and never need to. It reads
~/.testrelic/cli-config.json directly and writes through tr config set, so it's the same
config the CLI's own / commands use — the chat composer's / popup and the Command Center
never disagree.
Preferences, set from controls instead of typed commands:
- Engine — which assistant runs your tests: Automatic, TestRelic Cloud, Claude Code, Codex, or Ollama. See Modes & engines.
- Mode — the data boundary: on this computer, or Cloud.
- Permission — how much the assistant does before checking with you: Plan first, Ask every time, Write tests freely (recommended), or Full autopilot. Applies when it edits files on your computer.
- Chat display — show or collapse step-by-step tool activity in the transcript.
- Updates & memory — keep Studio auto-updated, let the assistant remember context
between sessions, and whether it follows your project's
TESTRELIC.mdrules.
All commands lists every command the chat understands, one click away: a setting jumps
to (and briefly highlights) its control above; an action — /init, /map, /heal,
/setup, /upload, /clear — starts a chat seeded with a friendly prompt instead of the
literal slash text.
A choice made here persists across restarts and takes effect the next time chat opens. Secrets like a Claude API key show only a redacted status, never the raw value, and can't be set from this UI — typing one into a native dialog would leak it into the process list.
Multi-session workbench
Alongside Chat, the sidebar's workbench row switches between three peer surfaces — Artifacts, Routines, and Dispatch — while the session list above it keeps every open session, grouped by project, one click away. Each session row shows a live status dot (idle, working, or offline) and can be renamed or closed in place.
- Routines — a saved task plus a cadence (hourly, every 6 hours, daily, weekly): a morning smoke suite, a nightly regression sweep. Toggle a routine on or off, run it on demand, and see its recent runs inline. A local scheduler fires each due routine automatically.
- Dispatch — a one-off background job: describe a task, hand it to the assistant, and walk away. The job list shows status at a glance; select one for its full task and result.
Routines and Dispatch both run through the same local job runner as an ordinary chat turn — a headless, auto-approving session with no one at the keyboard — so a scheduled or dispatched job honors the same Mode and engine as the rest of Studio, and every run (routine or one-off) shows up in Dispatch's job history.
Test Paths
A static navigation / action / assertion flow parsed straight from a test's source, independent of any run.
Run visualization
Steps, Console, Network, API, Artifacts, and Failure views sourced from your local
.testrelic-report/. A separate Traces tab opens a Playwright trace.zip in the
framework's own native Trace Viewer for a full step-by-step replay — screenshot, DOM, and
network for every step of a failed run.
Issue detection
Local, no-run heuristics: failures, flaky-by-retry, slow steps, console errors, and 4xx/5xx network calls.
| Type | Severity | Test | What happened | |
|---|---|---|---|---|
| Failure | High | applies promo code at checkout | Expected cart total to update, timed out after 5s | |
| Flaky | Medium | signs in with saved session | Passed 4/5 retries — intermittent redirect delay | |
| Slow | Low | GET /cart returns current items | Step exceeded 2× the suite median duration | |
| Console error | Medium | login flow on iOS simulator | Uncaught TypeError logged during render | |
| Network | High | shows validation error for expired card | 502 from /api/payments/validate |
Fix & re-run
Apply a suggested fix and re-run the test in place, with an edit → rerun → diff view. Fix
flows can drive your locally installed claude CLI directly, scoped to the repo, in addition
to the primary chat engine.
Apply a suggested fix and re-run in place — the diff view appears once the run finishes.
Terminal
An embedded terminal pane alongside chat.
Cloud Mode
Sign in once and the session is shared with the CLI. Signed in, the Runs tab replaces the local-only summary with real run history for the repo — resolved to your cloud repo by git identity (never auto-created, so just opening the tab can't mint an empty one) — showing outcome, commit, branch, CI provider, pass/fail/skip counts, duration, and a pass-rate trend for each run. Signed out, Runs shows your local report plus a "Sync to Cloud" prompt instead.
Insights adds org-wide aggregates — repo count, average pass rate, tests tracked — above the existing per-repo cards, plus a color-coded health bar and last-run time on every repo.
The status bar reflects the boundary at a glance: Cloud · synced when signed in, Local only when not.