The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use various system tools to analyze performance limits and hotspots.
Evidence: Mentions usage of uptime, nproc (Linux), sysctl (macOS), top, pidstat, node --cpu-prof, py-spy, perf, and strace -c.
Context: These tools are used within the context of investigating performance claims and identifying bottlenecks during the benchmarking process.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data that could potentially contain malicious instructions.
Ingestion points: The agent is directed to read measurement scripts provided by the user or found in the environment as part of the vetting process.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore or isolate embedded instructions within the analyzed scripts.
Capability inventory: The skill has the capability to execute the shell commands and profiling tools mentioned in the instructions.
Sanitization: There is no requirement or mention of sanitizing or validating the contents of the measurement scripts before they are read and analyzed.
[EXTERNAL_DOWNLOADS]: Fetches source metadata and content from a well-known repository.
Evidence: MIRROR.md indicates the skill content is sourced from the official Cursor plugins GitHub repository.

cursor-benchmark-checklist