The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill handles untrusted data from the store (e.g., product descriptions, reviews, customer notes) which could contain malicious instructions.
Ingestion points: Data enters the context through API operations such as list_variants and get_product as specified in the "Untrusted store text" section of SKILL.md.
Boundary markers: The skill includes explicit instructions for the agent to ignore instructions embedded in store text and to focus only on authorized task-relevant input.
Capability inventory: The skill possesses network capabilities via curl and MCP tools, and file-system read capabilities through bundled Node.js scripts.
Sanitization: Programmatic sanitization is not present; the skill relies on instructional boundaries for the AI agent.
[DYNAMIC_EXECUTION]: The skill uses local Node.js scripts to search the API specification and validate request schemas offline.
Evidence: SKILL.md defines Bash(node scripts/search_docs.mjs:*) and Bash(node scripts/validate.mjs:*) in the allowed-tools section. The search_docs.mjs script is dependency-free and processes a local gzipped JSON file.

a1-yandex-kit-catalog