The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions in SKILL.md reference the use of the Bash tool to execute local Python scripts (scripts/generate_schematic.py, scripts/compliance_checker.py, scripts/budget_calculator.py, scripts/deadline_tracker.py) for project management and figure generation. These scripts are invoked via the shell to process user-provided content.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its schematic generation workflow. In SKILL.md, the agent is instructed to run python scripts/generate_schematic.py "your diagram description" -o figures/output.png.
Ingestion points: The 'your diagram description' string is provided by the user at runtime.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to sanitize or escape shell metacharacters in this user-controlled string.
Capability inventory: The agent has access to the Bash tool, allowing it to execute arbitrary shell commands if the input string contains command injection payloads (e.g., ;, $(...), or backticks).
Sanitization: The skill lacks sanitization logic to prevent malicious user input from being interpreted as shell commands by the Bash tool.

research-grants