The Agent Skills Directory

[Command Execution] (HIGH): The scripts scripts/create_bug_report.sh and scripts/generate_test_cases.sh use the eval command to process user input, which is a major security risk.\n
Evidence: The function prompt_input in both scripts contains the line eval "$var_name=\"$input\"" where $input is read directly from the user/agent.\n
Risk: Shell metacharacters or command substitutions (e.g., $(whoami)) provided as input will be executed by the shell, leading to full system compromise.\n- [Indirect Prompt Injection] (HIGH): The skill provides an attack surface where untrusted data can lead to command execution.\n
Ingestion points: Interaction with the bash scripts create_bug_report.sh and generate_test_cases.sh which read data through the read command.\n
Boundary markers: None are used to delimit user input from the script logic or the generated markdown structure.\n
Capability inventory: The skill possesses the ability to execute shell commands (via the scripts) and write to the local filesystem.\n
Sanitization: There is no escaping or validation of the input before it is passed to eval or written to markdown files.

qa-test-planner