| name | run-generator-with-claude |
| description | Runs the AI test generator using Claude (Anthropic) instead of OpenAI. Use when the user wants to use Claude, Anthropic, or -p claude for test generation. |
Run generator with Claude
When to use
Use this skill when the user:
- Wants to generate API tests using Claude instead of OpenAI
- Says "use Claude", "run with Anthropic", or "use -p claude"
- Prefers Claude for the spec-to-test generation step
Workflow
-
Ensure API key is set.
The user must have ANTHROPIC_API_KEY set (e.g. export ANTHROPIC_API_KEY=your-key). Do not commit or log the key.
-
Run the generator with Claude provider.
From the ai-rest-assured-framework project root, pass the provider flag (e.g. -p claude) along with the spec path:
mvn exec:java -Dexec.args="--spec <path-to-spec> -p claude"
Example:
mvn exec:java -Dexec.args="--spec src/main/resources/sample-specs/amadeus-flight-most-traveled-destinations.json -p claude"
-
Same interactive flow.
The CLI will read the spec, call Claude for test cases, print them for review, and ask for approval before generating REST Assured + TestNG code.
-
Generated output.
Generated test classes are written under src/test/java/, typically in a generated/ package. Direct the user to open and review those files.
Notes
- If the CLI uses a different flag for provider (e.g.
--provider claude), use the project's actual flag. The intent is to run the generator with Anthropic/Claude as the LLM backend.
- OpenAI remains the default when no provider flag is given; this skill is for explicitly using Claude.
