The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements a persistent learning mechanism where it appends new discoveries to references/learnings.md. * Ingestion points: User requests regarding Spree Commerce platform features or troubleshooting processed in SKILL.md. * Boundary markers: None identified. The agent is instructed to append findings directly to the file without explicit delimiters. * Capability inventory: The skill specifies reading from references/learnings.md during initialization and appending to it in the final step of the instructions. * Sanitization: No explicit sanitization or validation of the content being appended is required, creating a surface for indirect prompt injection if malicious data is perceived as a valid 'learning'.
[EXTERNAL_DOWNLOADS]: The skill provides instructions and examples for downloading external tools and documentation. * Mentions installing agent skills from the spree/agent-skills GitHub repository via npx. * References an external MCP server for documentation at https://spreecommerce.org/docs/mcp. * Includes commands to install related vendor tools from sales-skills/sales.

sales-spree