The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied messages by base64-encoding them and passing the result as arguments to internal validation and logging scripts.
Ingestion points: The user's most recent prompt is ingested, base64-encoded, and passed as the --user-prompt-base64 argument to scripts/validate.mjs and the --comment-base64 argument to scripts/log_feedback.mjs.
Boundary markers: Instructions in SKILL.md direct the agent to enclose the base64 string in single quotes within the shell command to isolate it from the execution context.
Capability inventory: The skill uses the bash tool to execute local Node.js and shell scripts. These scripts have network access to shopify.dev for search and telemetry.
Sanitization: The use of base64 encoding prevents the user prompt from being interpreted as shell commands or script logic during the handover to the helper tools.
[DATA_EXFILTRATION]: The skill implements a telemetry system that transmits tool invocation metadata and the user's prompt to Shopify's infrastructure.
Evidence: The PostToolUse hook in SKILL.md executes scripts/track-telemetry.sh, which collects session details and user prompts to send to https://shopify.dev/mcp/usage via curl.
Context: The destination domain shopify.dev belongs to the skill's author, and the telemetry behavior is disclosed in the skill's privacy notices.
[COMMAND_EXECUTION]: The agent is instructed to use the bash tool to run specific helper scripts for documentation searching and code validation.
Evidence: Mandatory steps in SKILL.md require the execution of scripts/search_docs.mjs and scripts/validate.mjs.
[EXTERNAL_DOWNLOADS]: The skill specifies several Node.js dependencies in package.json hosted on the NPM registry.
Evidence: Dependencies include @shopify/theme-check-common, @shopify/theme-check-docs-updater, and @shopify/theme-check-node. These are official packages from the vendor.

shopify-liquid