The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests storefront plan and brand documentation which acts as an untrusted input surface.
Ingestion points: The skill reads external files including brand.md, design.md, persona.md, products.md, and plan documents in SKILL.md.
Boundary markers: No explicit markers or instructions to ignore instructions embedded within these files are provided.
Capability inventory: The skill can discover tools, generate assets using lexsis_drafts.asset_generate, and perform imports via lexsis_asset_import.import.
Sanitization: No sanitization or data validation steps are identified in the instructions.
[EXTERNAL_DOWNLOADS]: The skill connects to an external storefront management service via MCP.
Evidence: agents/openai.yaml specifies a connection to the lexsis-ai MCP server hosted at https://mcp.trylexsis.com/mcp.

visualize-page