The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill demonstrates how to implement a simple JIT compiler using mmap to allocate memory with executable permissions (PROT_EXEC) and execute machine code at runtime. The instructions include specific notes for macOS Apple Silicon regarding mprotect and MAP_JIT, indicating an awareness of platform-specific security requirements for executable memory.
[COMMAND_EXECUTION]: The skill suggests using the Linux perf tool for profiling the interpreter loop performance. Examples include perf record and perf report to identify branch mispredictions and dispatch overhead.
[INDIRECT_PROMPT_INJECTION]: The skill focuses on building runtimes that process external bytecode. While this creates a data ingestion surface, the skill explicitly references related fuzzing skills to secure the bytecode parser and loader against malicious input.

interpreters