The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied source code and performance logs to identify bottlenecks, which represents a surface for indirect instructions embedded in untrusted data. \n
Ingestion points: User-provided source code, script files, and profiling outputs (SKILL.md). \n
Boundary markers: None defined within the skill instructions. \n
Capability inventory: References shell command execution (strace, iotop, bash -x) and package installation (pip install) (SKILL.md). \n
Sanitization: No explicit input validation or sanitization logic is provided. \n- [EXTERNAL_DOWNLOADS]: Recommends the installation of well-known Python profiling and analysis tools such as snakeviz, line-profiler, and memory-profiler via standard package managers. \n- [COMMAND_EXECUTION]: Lists various system monitoring and profiling commands, including strace, iotop, htop, top, and hyperfine, for performance analysis. \n- [DYNAMIC_EXECUTION]: Suggests the use of numba's JIT compilation and Cython as techniques for optimizing execution speed at runtime.

performance-optimizer