The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions command the agent to run Python scripts (e.g., Phase 1-3) with arguments derived from user input or external sources (e.g., <competitor_blog_url>, <keywords>, <linkedin_url>). This direct interpolation into shell commands presents a significant risk of command injection if the input strings contain shell metacharacters.
[PERSISTENCE]: The skill documentation explicitly provides a 'Scheduling' section that instructs the user to implement a cron job to maintain recurring execution of the skill across sessions.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill retrieves content from external, third-party sources including arbitrary blog URLs, LinkedIn profile posts, and Twitter/X feeds.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to distinguish between the skill's own instructions and the potentially malicious data retrieved from external scrapes.
Capability inventory: The skill has access to shell command execution (via the scraper scripts) and the ability to write files to the local system (saving configurations and intelligence reports).
Sanitization: The skill lacks any defined sanitization, filtering, or validation process for the data ingested from the open web before it is analyzed and synthesized into the final output.

competitor-content-tracker