The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using local Python scripts where user-provided input is directly interpolated into the command string.
Evidence: python scripts/dfs_client.py serp --keyword "<pillar-keyword>" --limit 10 and python scripts/report_publish.py <report>.md.
Security Risk: A malicious user could provide a keyword or filename containing shell metacharacters (e.g., "; rm -rf /") to execute unintended commands on the system, as the skill lacks explicit instructions for sanitization or parameterization.
[INDIRECT_PROMPT_INJECTION]: The skill fetches content from external websites to analyze section coverage and link patterns, which creates a surface for adversarial instructions to enter the agent's context.
Ingestion points: External content is retrieved using the webfetch tool from top-ranking URLs discovered in search results in the "Data pulls" section.
Capability inventory: The skill possesses the capability to execute shell scripts (Python) and write files to the local file system as defined in the "Process" and "Output location" sections.
Boundary markers: The instructions do not define specific delimiters or explicit warnings for the agent to ignore or isolate instructions potentially hidden within the fetched web content.
Sanitization: There are no sanitization, filtering, or validation steps defined for the data retrieved via webfetch before it is used to determine the page's structure, scope, and word count targets.

pillar-page-designer