Available
Feature · Agentic Authoring

Author structured scenarios with the agents you already use

Claude Code, Codex, Cursor, and SolidX Agent author TestHub scenarios through the SolidX testing skill and MCP. The output is structured, reviewable, and runs the same way every time.

One canonical vocabulary

A small, explicit operation set every scenario shares

Agents do not invent test syntax. Every scenario is composed from the same vocabulary, which is exactly what the TestHub runner executes and the portal renders in human-readable steps.

api.requestSend an HTTP request and optionally save the response for later steps.
assert.httpStatusAssert the response status code.
assert.jsonPathAssert a value at a JSON path in the response.
assert.containsAssert the response contains expected text.
ui.gotoNavigate the browser to a page.
ui.clickClick an element by selector or visible text.
ui.fillType into an input.
ui.expectVisible / ui.expectTextAssert what the user should see.
util.requireDeclare dependencies between scenarios in a flow.

URLs are templated with ${env:...} values and responses are reused across steps with ${res:...} references, so scenarios never hard-code hosts or duplicate literal values.

Why it works

Agent speed, engineering discipline

Structured by default

Agents produce scenarios in one canonical vocabulary with stable IDs, tags, parameters, timeouts, and retries. Reviewable diffs, not free-form scripts.

Grounded, not guessed

Agents inspect the connected repository through TestHub's repository intelligence layer: maps, outlines, search, and bounded reads. Routes and selectors come from source, not imagination.

The file is the truth

Scenarios live in a structured JSON definition that is the single source of truth. The portal watches the file and refreshes live on every change.

Reusable by design

Environment templating, saved response reuse, fixtures, and inter-scenario dependencies keep suites DRY as they grow.

Governed access

Agents get application-scoped, read-only repository access. No Git tokens, no absolute paths, no shell access, and every call is audited.

Runs the same everywhere

Agent-authored scenarios run in the portal, locally, or in CI through solidctl with identical, deterministic behavior.

On the roadmap

Agentic authoring, directly inside the portal

Today your agents author through MCP using the tools and skills they already have. Next, TestHub is bringing the same agentic authoring loop into the portal itself, so teams can go from intent to reviewed scenario without leaving the product.

In-portal agentic authoring joins the existing flow: describe the behavior, review the grounded draft, and accept it into your scenario library, all inside TestHub.

Explore all TestHub capabilities · Currently viewing: Agentic Authoring

Watch an agent author a scenario for your app

Request a walkthrough and we will author, review, and run a scenario against your application live.

Part of SolidX