The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various system commands including mvn test, git diff, docker info, and platform-specific file openers (open, xdg-open, start). These are standard for development workflows but represent a capability to interact with the host environment.
[DYNAMIC_EXECUTION]: A Python script is embedded within SKILL.md and executed via a shell heredoc (`python3
<<'PY'). The script is used to parse a local HTML coverage report (target/coverage-report/report.html`) generated by the test suite.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted data from the project's source files and generated test reports.
Ingestion points: The agent reads BPMN and DMN files from src/main/resources/processes/ and parses the HTML coverage report at target/coverage-report/report.html.
Boundary markers: No specific delimiters or "ignore instructions" warnings are used when processing these files.
Capability inventory: The agent can execute shell commands (mvn, git) and dynamic Python scripts based on the results of the ingested data.
Sanitization: The Python parsing script uses regular expressions and json.loads to extract data from the HTML report but does not explicitly sanitize the content for malicious instructions.
[EXTERNAL_DOWNLOADS]: The skill relies on standard package managers and tools (Maven, Docker, GitHub Actions) to download dependencies and container images. These operations target well-known services like Maven Central and Docker Hub, which is standard practice for the intended Java development use case.

camunda-process-test