The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's instructions (Step 6) direct the agent to execute local Python scripts using relative paths that navigate outside the skill's own directory structure.
Evidence: The commands python ..\\skill-creator-advanced\\scripts\\format_check.py . and python ..\\skill-creator-advanced\\scripts\\quick_validate.py . in SKILL.md.
Risk: This pattern assumes a specific local file environment and attempts to execute code from a parent directory, which could be exploited for unauthorized local code execution if the agent's environment lacks strict sandboxing.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and process content from external web sources, creating a surface for indirect prompt injection attacks.
Ingestion points: Web search results retrieved in Step 2 of the workflow.
Boundary markers: Absent. The instructions do not specify the use of delimiters or 'ignore instructions' markers to isolate external content.
Capability inventory: The agent is granted command execution capabilities via the instructions to run Python scripts.
Sanitization: Absent. There are no mechanisms described to sanitize or filter potential instructions embedded in retrieved web content.
[DYNAMIC_EXECUTION]: The skill facilitates dynamic execution by providing hardcoded command-line calls to local Python scripts at runtime.
Evidence: Step 6 in SKILL.md explicitly lists commands to invoke the Python interpreter on specific file paths.

concept-alignment