The Agent Skills Directory

[SAFE]: The skill's primary focus is providing architectural guidance and analysis tools for context engineering. The included Python scripts perform passive analysis on conversation logs.
[INDIRECT_PROMPT_INJECTION]: The context_analyzer.py and compression_evaluator.py scripts are designed to ingest conversation histories from external JSON files for analysis. While processing untrusted data provides a surface for indirect prompt injection, the scripts themselves do not possess dangerous capabilities such as network communication, file writing, or command execution, ensuring that any malicious input in the data would only result in inaccurate analysis metrics rather than system compromise.

context-engineering