The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches live HTML content from external, user-provided domains to verify audit findings. This creates a surface for indirect prompt injection if the audited site contains malicious instructions hidden in its code.
Ingestion points: Workflow step 5 involves fetching live page HTML; step 3 reads data from get_audit_issues and get_audit_pages within SKILL.md.
Boundary markers: There are no explicit delimiters or instructions to the agent to ignore content that looks like commands within the fetched HTML.
Capability inventory: The skill has the ability to write to the project context (update_project_context), add key pages (addKeyPages), append to a research log (appendResearchLog), and generate HTML reports via the seo-report tool (SKILL.md).
Sanitization: No specific sanitization, escaping, or filtering of external HTML content is described before the agent processes it or includes it in the final report.
[COMMAND_EXECUTION]: The skill invokes several MCP tools (run_site_audit, get_backlinks_overview, research_keywords, get_domain_overview) to interact with the OpenSEO service and generate its reports. This is the intended primary functionality of the skill.

seo-audit