The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill resolves the path to its utility scripts (<suite-root>) dynamically at runtime using environment variables (CLAUDE_PLUGIN_ROOT) or local configuration files (.seo-suite-runtime.json), then executes Python scripts from this computed location.
[INDIRECT_PROMPT_INJECTION]: The skill processes external web data such as the DOM and accessibility tree to audit agent behavior, which creates a surface where malicious instructions hidden in untrusted web content could influence the agent.
Ingestion points: Processes web page DOM and accessibility trees during audits (SKILL.md, references/agent-task-protocol.md).
Boundary markers: Absent; no specific delimiters or warnings are defined to isolate external web content.
Capability inventory: Executes local Python scripts for validation, site building, and work indexing (SKILL.md, references/markdown-delivery-contract.md).
Sanitization: Absent; the skill uses deterministic verification but does not describe explicit sanitization for web data inputs.
[COMMAND_EXECUTION]: The skill instructs the agent to run various Python command-line tools for auditing and documentation management, including validate_seo_findings.py, manage_markdown_site.py, and manage_markdown_work.py.

seo-agentic