UiPath Takes Coding Agents GA and Previews a Dark Testing Factory
UiPath said on September 23 that **UiPath for Coding Agents is generally available** across its platform. At its FUSION 2026 event, the company also showed building blocks for what it calls a **dark testing factory**: a proposed system in which agents and deterministic automation handle more of the continuous testing loop. The two announcements point to a wider software factory trend, but they are at different stages of release. [1]
UiPath first introduced its coding-agent integration in May. The September milestone is general availability, not the product's first appearance. Its testing factory is further out: UiPath calls it a longer-term vision and says some components are available today. [1] [3] [4]
What developers can use now
UiPath for Coding Agents is designed to work with the coding agent a team already uses, including Claude Code, OpenAI Codex, Cursor, and other agents that follow the AGENTS.md format. UiPath says its CLI and open-source agent skills give those agents a way to scaffold cases, workflows, API automations, and process orchestrations, then validate, test, package, deploy, and operate them through the platform. This is an integration with UiPath's automation stack, not a new foundation coding model. [1] [2]
The product page describes local validation, unit tests, and workflow analyzer checks as feedback that an agent can iterate against. Packaging and promotion follow UiPath's existing governance paths, with identity, role-based access controls, audit trails, and policy applied to the resulting automation. UiPath also says that approved enterprise process specifications can inform what a coding agent builds, with evaluations gating deployment. Those are vendor descriptions of the workflow, not independently measured quality outcomes. [1] [2]
| Item | Status reported by UiPath on September 23 |
| --- | --- |
| UiPath for Coding Agents | Generally available across the platform |
| Test Cloud agentic testing capabilities | Available capabilities and FUSION demonstrations |
| Full dark testing factory | Longer-term vision; building blocks previewed |
| External Playwright tests in Test Manager | Integration under development |
| Pricing | No standalone price published in the cited product materials; trial offered |
The UiPath product page offers a free trial and says subscriptions to third-party AI models must be purchased separately. It does not give a standalone price for UiPath for Coding Agents, so teams should check their UiPath plan and model costs before estimating deployment expense. [2]
The proposed testing factory
UiPath describes the dark testing factory as a quality system in which agents and scripts increasingly understand incoming changes, assess risk, design and execute tests, analyze results, maintain test assets and environments, and route exceptions. The aim is to automate more of the operating loop, not merely generate individual test cases. People still define goals, provide context, set boundaries, and decide ambiguous or consequential cases. [1] [3]
The company's Test Cloud article gives concrete examples of current direction: autonomous execution of a manual test through UiPath Autopilot, exploratory testing driven by a tester's mission, and semantic or AI-based verification for applications whose outputs vary between runs. It also describes an AI Trust Layer for activity auditing, spend controls, and masking personally identifiable information. These features need to be evaluated on a team's own applications; the announcement does not supply an independent comparison against existing test pipelines. [3]
UiPath also says it is working toward running Playwright tests created outside UiPath through Test Manager and adding self-healing for shifted selectors. That work is presented as an integration goal, not a generally available capability. [3]
Why the distinction matters
The relevant developer question is how coding agents' output moves through validation, review, testing, and deployment. UiPath's GA product addresses the build and governance side for automations on its platform. The dark testing factory describes a more autonomous quality loop that could eventually respond to software changes at a similar pace. Both depend on reliable evidence, permissions, and human decisions at the points where automation is uncertain. [1] [2] [3]
This news is therefore a concrete sign of the software factory trend, not proof that a fully autonomous factory has arrived. For teams considering it, the useful near-term test is narrow: let a coding agent create one UiPath automation, run the platform's validation and tests, inspect the audit trail and approvals, and compare the result with the team's existing process. The broader testing vision should be judged as its components ship and their effect on defect detection, review load, and operating cost becomes measurable.