The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external data from requirements and Figma designs, which could contain malicious instructions designed to subvert agent behavior.\n
Ingestion points: Requirement descriptions in user prompts and Figma design metadata accessed through the Figma MCP tool.\n
Boundary markers: None. The skill does not use specific delimiters or instructions to isolate external data from its operational logic.\n
Capability inventory: The skill can create and write to markdown files and execute internal bash utility scripts.\n
Sanitization: None. Inputs are interpolated directly into documentation templates.\n- [DYNAMIC_EXECUTION]: The skill includes bash scripts (scripts/generate_test_cases.sh and scripts/create_bug_report.sh) that use the eval command to process user-supplied input.\n
Evidence: Both scripts use the eval command for dynamic variable assignment in the prompt_input function.\n
Risk: The use of eval with unsanitized user input allows for command injection if the input contains shell metacharacters or subshell execution markers.\n- [COMMAND_EXECUTION]: The skill relies on the execution of internal bash scripts that lack sufficient input validation.\n
Evidence: The scripts generate_test_cases.sh and create_bug_report.sh take string inputs that are processed unsafely, potentially allowing an attacker to execute arbitrary commands if they can control the inputs passed to the script.

qa-test-planner