The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes instructions that mandate a grounding step using a specific documentation search tool to ensure accuracy. These instructions guide the agent's workflow for its intended purpose and do not attempt to override safety filters or bypass ethical constraints.
[EXTERNAL_DOWNLOADS]: The scripts docs-search.mjs, graphql-schemata.mjs, and openApi-schemata.mjs perform HTTP GET requests to docs.commercetools.com. These operations fetch documentation and API specifications from the vendor's official infrastructure to provide context for the agent.
[INDIRECT_PROMPT_INJECTION]: The skill implements a grounding mechanism where the agent ingests content from a remote API.
Ingestion points: scripts/docs-search.mjs (fetches data from docs.commercetools.com).
Boundary markers: Present; the search script formats the output with markdown headers (## Result) and labels (**Content:**) to help the agent distinguish search results from instructions.
Capability inventory: The skill includes Node.js scripts that perform network requests (fetch) and print to stdout for the agent to read.
Sanitization: The fetched content is not sanitized for instructions, but the risk is mitigated by the fact that content originates from the official vendor documentation site for the product being integrated.

commercetools-checkout