The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its website audit functionality.
Ingestion points: The scripts/seo_audit.py script fetches content (titles, meta descriptions, and H1 tags) from arbitrary URLs provided by the user.
Boundary markers: The output of the script is presented to the agent without explicit delimiters or instructions to ignore embedded commands within the fetched HTML content.
Capability inventory: The agent can execute local Python scripts, perform network requests, and search the web, which could be exploited if the agent follows instructions found on a malicious website being audited.
Sanitization: The script uses regular expressions to extract text from HTML tags, which provides some filtering but does not prevent semantic instructions within that text from being processed by the LLM.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to external services as part of its core functionality.
Well-known services: Scripts such as scripts/dataforseo_api.py and scripts/keyword_research.py fetch data from api.dataforseo.com. This is a well-known industry service for SEO data and is documented neutrally.
Audit requests: The scripts/seo_audit.py script performs GET requests to user-specified URLs to analyze their SEO meta tags.
[CREDENTIALS_UNSAFE]: The skill handles API credentials safely.
Credential Management: The scripts/credential.py and scripts/dataforseo_api.py files retrieve API logins and passwords from the DATAFORSEO_LOGIN and DATAFORSEO_PASSWORD environment variables, which is the recommended practice for local agent skills.

seo-geo