prd-test-writer
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The generated review HTML templates in the
samples/andassets/directories include references to the Mermaid.js library hosted on the JSDelivr CDN (https://cdn.jsdelivr.net/npm/mermaid@11/dist/mermaid.esm.min.mjs). JSDelivr is a well-known and trusted service for delivering open-source frontend assets. - [INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting local project source code during its 'Autonomous Code Reading' phase (Stage 1) to inform the generation of requirements and test specifications. While this represents a potential attack surface for indirect prompt injection from malicious source files, the skill implements a multi-stage review process (Stage 5) utilizing independent sub-agents to verify the integrity and honesty of the generated documents.
- [COMMAND_EXECUTION]: The test case samples and templates reference shell commands for running tests (e.g.,
npm run smoke:provider). These are documented as instructions for the developer to execute within a local environment to verify implementation, rather than commands for the agent to execute autonomously or silently on the host system.
Audit Metadata