The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill uses importlib and spec_from_file_location in its __init__.py to dynamically load its own internal scripts (bm25_retriever.py and jev_rerank.py). This is a standard pattern for multi-file plugins in the Hermes Agent environment.
[DATA_EXFILTRATION]: When the optional jev reranking feature is explicitly enabled by the user via environment variables, the skill sends truncated user queries and skill descriptions to api.typesafe.ai or openrouter.ai. This behavior is documented as a core feature for semantic reranking and includes character caps to limit data exposure.
[COMMAND_EXECUTION]: The package's test suite (tests/test_stdlib_parity.py) utilizes subprocess.run to verify that the runtime environment is free of specific heavy numeric dependencies. This operation is confined to the development and testing context and is not part of the skill's primary execution path.
[PROMPT_INJECTION]: The skill implements a neutralize_query function in jev_rerank.py that utilizes regex patterns to strip common imperative instruction-bypass phrases from user input before processing. This is a proactive security measure designed to mitigate prompt injection risks.

skill-retrieval