The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes npx artillery to run load tests. This command fetches the Artillery package from the npm registry, which is a standard and well-known service for Node.js package distribution.
[COMMAND_EXECUTION]: The skill generates and executes shell commands for benchmarking tools including k6, artillery, and ab (Apache Benchmark). These commands are necessary for the skill's primary purpose of measuring system performance.
[INDIRECT_PROMPT_INJECTION]: The skill interacts with external configuration and data files, which presents a surface for indirect prompt injection.
Ingestion points: Reads from an approved hosts configuration file and synthetic test data (e.g., users.csv) as described in the SKILL.md security section.
Boundary markers: The instructions mandate restricting tools to whitelisted domains and IPs, effectively limiting the scope of execution.
Capability inventory: Uses network-capable CLI tools (k6, artillery, ab) to perform stress and load testing.
Sanitization: Includes mandatory guardrails such as a downstream health-check loop and an automated kill switch to stop tests if error rates exceed 5%.

system-benchmark