The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes master prompts designed to direct an LLM to act as a senior QA engineer. These prompts use role-play ("You are a senior QA engineer with 20+ years of experience") and directive language ("CRITICAL INSTRUCTIONS", "MANDATORY RULES") to ensure the agent follows the structured testing workflow. This is consistent with the skill's primary purpose of providing an autonomous testing framework.
[COMMAND_EXECUTION]: The skill facilitates command execution through initialization scripts (init_qa_project.py) and documentation that guides users and agents in setting up development environments and running test cases. These commands (e.g., git clone, pnpm install, ccpm install) are standard for software development and QA activities.
[DATA_EXFILTRATION]: There is no evidence of unauthorized data exposure or exfiltration. The skill encourages best practices for secret management, such as using .env files for local environment variables, and focuses on analyzing project-local quality metrics.
[INDIRECT_PROMPT_INJECTION]: As an autonomous testing tool, the skill has an inherent attack surface for indirect prompt injection. It is designed to ingest and process untrusted project data, including source code and test specifications. While the prompts include boundary instructions ("DO NOT deviate from documented test cases") to minimize this risk, the execution of commands found within test documents represents a standard capability associated with this use case.

qa-expert