The Agent Skills Directory

[DATA_EXFILTRATION]: The skill's scripts perform network operations to an external domain (api.dataforseo.com) and arbitrary target URLs during audits. While this is the intended purpose, it involves sending API credentials over the network to a non-whitelisted service.
Evidence: scripts/dataforseo_api.py sends DATAFORSEO_LOGIN and DATAFORSEO_PASSWORD via Basic Authentication to https://api.dataforseo.com/v3.
Evidence: scripts/seo_audit.py can be instructed to fetch the source code of any provided URL using urllib.request.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the DataForSEO API responses and from external websites during SEO audits. This content could potentially contain malicious instructions meant to influence the AI agent's behavior if processed without adequate delimiters.
Ingestion points: api_post in scripts/dataforseo_api.py and fetch_url in scripts/seo_audit.py.
Boundary markers: The instructions do not define strict delimiters or 'ignore' instructions for the data fetched from these tools.
Capability inventory: The skill has network access and file system write capabilities (for saving drafts), which increases the potential impact of an injection.
Sanitization: Data is parsed as JSON or via regex, but the content is not explicitly sanitized against secondary prompt injection attacks.

seo-and-aeo-strategy