cawplan-testcase-import
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes test case data from external sources, such as product portals and requirement links, which represent a significant ingestion surface for untrusted content.
- Ingestion points: Data source defined in the
argument-hintas product portal links or requirement links (SKILL.md). - Boundary markers: The skill mandates a 'preview-first' workflow and an explicit 'execute --confirm' human-in-the-loop gate before final execution (SKILL.md).
- Capability inventory: The agent can execute shell commands, perform network operations via the
qa-insightsCLI tool, and run local Node.js scripts to transform data (SKILL.md, references/import-rules.md). - Sanitization: The conversion script (
scripts/convert_generate_to_import.js) performs schema validation and checks field requirements, but does not implement filtering or sanitization for prompt injection patterns within the text content of the test cases. - [COMMAND_EXECUTION]: The skill uses the
Bashtool to execute various CLI commands includingqa-insights testrail,jobs, andnodeto manage the import workflow and interact with external APIs (SKILL.md).
Audit Metadata