The Agent Skills Directory

External content retrieval: The skill performs network operations using curl or wget to fetch a JSON catalog and markdown instructions from official repositories. This allows the agent to access up-to-date product information rather than relying on preloaded data.
Dynamic instruction loading: Once relevant matches are identified, the skill fetches and follows instructions from remote entrypoint URLs. This represents a dynamic execution pattern where the agent's behavior is guided by instructions retrieved at runtime.
Command-line tool usage: The workflow utilizes system utilities such as jq, node, and PowerShell to filter and process the retrieved catalog. The skill includes explicit security guidelines, such as mandating secure TLS connections and forbidding the use of insecure flags like -k or --insecure during these operations.
Data ingestion surface: The skill ingests and processes external content (JSON and Markdown) from the google/skills repository.
Ingestion points: Content is retrieved via curl, wget, or node in SKILL.md (Step 1 and Step 4).
Boundary markers: The instructions do not specify unique delimiters for the fetched content, but they provide strict schema validation (checking for the skills array) to ensure data integrity.
Capability inventory: The skill has capabilities for network access and the execution of shell commands for data processing.
Sanitization: The skill emphasizes transport-layer security (SSL/TLS verification) to prevent tampering during ingestion.
Runtime script execution: The skill provides a fallback mechanism using node -e to execute a small script for fetching data if other tools are unavailable. This is a controlled use of dynamic execution for the specific purpose of content retrieval from a trusted source.

finding-google-skills