The Agent Skills Directory

[COMMAND_EXECUTION]: The helper scripts scripts/create_bug_report.sh and scripts/generate_test_cases.sh are vulnerable to arbitrary shell command injection. The scripts use the read command to get user input and then pass it directly into an eval statement (e.g., eval "$var_name=\"$input\""). An attacker could provide a payload like "; whoami; # to execute unintended commands in the user's shell environment.
[DYNAMIC_EXECUTION]: The use of eval in the provided shell scripts to dynamically assign variables from unvalidated input is a high-risk pattern. This allows the execution of code generated at runtime based on external data.
[INDIRECT_PROMPT_INJECTION]: The skill integrates with external Figma design files to retrieve specifications. This data is untrusted and could contain malicious instructions designed to influence the agent's behavior during test case generation or bug reporting.
Ingestion points: Figma MCP server responses triggered by user-provided URLs (referenced in SKILL.md and references/figma_validation.md).
Boundary markers: None present; the templates and guides do not include delimiters or instructions to ignore embedded content in the design files.
Capability inventory: The skill allows writing markdown files to the local filesystem and executing interactive terminal scripts.
Sanitization: No sanitization or validation of the content retrieved from Figma is performed before it is used to populate test cases or bug reports.

qa-test-planner