Category ยท 7 integrations
AI model providers integrations
Choose the model provider your agents run on and use your own account and API key.
Connecting a model provider decides which models power your agents and whose account is billed. It is different from connecting ChatGPT or Claude to TestingOps, which is a connection guide.
Compare
How these tools differ
| Tool | What TestingOps does with it | Connection |
|---|---|---|
| Anthropic Claude | Agents switch to the Claude adapter and run on your Anthropic key; the Model field suggests the models Anthropic lists for that key. | Model provider Available |
| OpenAI | Agents switch to the Codex adapter on your OpenAI key; the key shares one variable with the OpenAI-compatible providers. | Model provider Available |
| Google Gemini | Covers two roles: agents run on your Gemini key, and Gemini CLI can call TestingOps over MCP. | Model provider, Connection guide Available |
| Azure OpenAI | Agents run through Codex against your own Azure resource; needs the resource's openai/v1 endpoint and your deployment name. | Model provider Available with limits |
| xAI Grok | Agents run on the Grok adapter with your xAI key; only with your own key, and Grok agents receive no MCP tools. | Model provider Available with limits |
| Qwen | Agents run Qwen through OpenCode on your Model Studio key; shares one OpenAI-compatible slot, and TestingOps' own AI features do not use it. | Model provider Available with limits |
| Kimi | Agents run Kimi through OpenCode on your Moonshot key at api.moonshot.ai; shares the OpenAI-compatible slot with Qwen and others. | Model provider Available with limits |
Integrations in this category
Anthropic Claude
Agents on your projects run Claude through your own Anthropic account and API key.
Model providerAvailableOpenAI
Agents on your projects run OpenAI models through the Codex adapter with your OpenAI API key.
Model providerAvailableGoogle Gemini
Agents can run on your Gemini API key, and Gemini CLI can use TestingOps over MCP: two separate connections.
Model providerConnection guideAvailableAzure OpenAI
Agents on your projects run on your own Azure OpenAI deployment through the Codex adapter.
Model providerAvailable with limitsxAI Grok
Agents on your projects run Grok with your xAI API key; Grok agents do not get MCP tools.
Model providerAvailable with limitsQwen
Agents on your projects run Qwen through OpenCode with your Alibaba Cloud Model Studio key.
Model providerAvailable with limitsKimi
Agents on your projects run Kimi models through OpenCode with your Moonshot AI key.
Model providerAvailable with limits
Other categories
Issue trackers, wikis and code
Requirements, work items, documentation and code changes are where test design starts. These connections give TestingOps agents that context and, where supported, a place to report back.
Test management
Keep your test management tool as the system of record. TestingOps can send cases with their steps, bring existing cases into its knowledge base and report run results where the tool supports it.
Design tools
Screens, flows and workshop boards describe the intended behaviour before any code exists. These connections bring that material into the knowledge base that agents write test cases from.
Test execution
TestingOps runs test cases on an execution provider and brings back results, steps and evidence. Which providers a workspace can use is decided by its execution policy.
Low-code platforms
Low-code platforms keep the application as a model: data, screens and logic in a structured form. Reading that model lets agents design tests around what the application actually contains.
Process models (BPMN)
A business process model already describes the paths a case can take. TestingOps turns BPMN 2.0 elements into bounded test scenario candidates for review.
App markets
Store reviews show where real users struggle. TestingOps turns them into insights for the knowledge base: themes, version-level signals and test ideas.
MCP tools for agents
These connections run in one direction: your TestingOps agents call an external tool through the Model Context Protocol (MCP) when a task needs it.
Connection guides
Use TestingOps from the apps you already work in. These are setup guides: you connect the app to TestingOps, and the app calls TestingOps.
Private networks
Many test environments are only reachable inside a company network. These options let test runs reach them without opening the environment to the internet.