The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill executes an embedded Python script via a shell heredoc to parse the HTML coverage report generated by Maven. This script is statically defined in the instructions and is used to extract JSON coverage data for analysis by the agent.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data including BPMN models, DMN tables, and generated HTML coverage reports. This creates a surface where content embedded in process models (such as element IDs or descriptions) could influence the agent's behavior when it reviews coverage results.
Ingestion points: The agent reads target/coverage-report/report.html (via the Python helper) and local *.bpmn, *.dmn, and *.test.json files.
Boundary markers: No explicit boundary markers or instruction-override warnings are used when processing the data extracted from reports.
Capability inventory: The skill can execute shell commands (mvn, python3, git), read/write files (test scenarios), and access the network via Maven for dependency resolution.
Sanitization: The Python script uses standard JSON parsing and regex, which prevents direct shell injection from the report data, but the resulting element IDs are passed back to the LLM context without further filtering.
[EXTERNAL_DOWNLOADS]: The skill uses Maven (mvn test) to run tests, which involves downloading Java dependencies and plugins from official registries. These operations are performed via standard build tooling.
[COMMAND_EXECUTION]: The skill executes several CLI tools including mvn, git, python3, and Docker commands to manage the test environment and lifecycle. These commands are necessary for the primary purpose of process testing.

camunda-process-test