The Agent Skills Directory

[SAFE]: No security concerns were identified during the analysis of the instructions or the accompanying scripts.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill does not perform any network operations, use any communication tools, or access sensitive local files. All data processing is local to the agent's execution environment.
[REMOTE_CODE_EXECUTION]: The provided Python scripts (script.py and skill.py) implement medical logic using standard built-in libraries. There is no usage of eval(), exec(), or subprocess calls that could facilitate arbitrary code execution.
[PROMPT_INJECTION]: The instructions in SKILL.md and instructions.md are purely informational and do not contain any patterns intended to bypass safety guardrails or override system instructions.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits a safe ingestion surface.
Ingestion points: Data enters via the run_acls_protocol function in script.py and skill.py through arguments like rhythm_str and symptoms.
Boundary markers: Not explicitly required as the skill lacks dangerous capabilities.
Capability inventory: No network access, file system writes, or command execution capabilities are present across any script.
Sanitization: The rhythm_str input is strictly validated against a predefined CardiacRhythm Enum, preventing the processing of arbitrary instruction strings through that vector.

acls-protocol-assistant