The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides an audit log analyzer script (scripts/audit-log-analyzer.py) that processes external data files, representing a potential attack surface for indirect prompt injection where malicious log entries could attempt to influence the agent's compliance reporting or subsequent actions.
Ingestion points: The load_audit_log method in scripts/audit-log-analyzer.py reads from user-supplied log files in JSON and CSV formats.
Boundary markers: There are no explicit boundary markers or instructions to the agent to disregard embedded content within the log entries during analysis.
Capability inventory: The skill possesses file-writing capabilities through scripts/audit-log-analyzer.py, which generates a compliance report (compliance_report.json) using json.dump.
Sanitization: The script performs standard structure-based parsing (JSON/CSV) but does not validate or sanitize the string content of the logs against instructional patterns.
[DATA_EXPOSURE]: The main instruction file SKILL.md contains an absolute file path reference (/home/william/git/standards/docs/compliance/healthtech/implementation-guide.md) which exposes the author's local filesystem structure and creates a non-portable link that may not function in different execution environments.

healthtech-hipaa