The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external websites (competitors, live site crawling, and search results) while maintaining the capability to modify the repository's source code and configuration. This creates a surface where malicious instructions embedded in external content could influence the agent's behavior.
Ingestion points: Crawling the live site (Daily step 1), fetching competitor sitemaps (Weekly step 5), and reading external search results (Daily step 1).
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content within the fetched data.
Capability inventory: The agent is authorized to create and edit files, routes, blog posts, and redirects within the repository.
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from external URLs before it is processed.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to retrieve and follow instructions from external third-party domains that are not part of the established trusted vendor list.
Evidence: The skill explicitly directs the agent to "Read https://linkdr.com/SKILL.md and follow it" and use "https://aiseotracker.com/docs".
[COMMAND_EXECUTION]: The skill requires the agent to run project-specific maintenance and build commands, which could be leveraged if the agent's instructions are compromised.
Evidence: Instructions to "run the repo's lint, type check and build" before shipping changes.

seo-autopilot