The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data, creating a surface for indirect prompt injection where malicious instructions could be embedded in data fields to influence agent behavior.
Ingestion points: The agent is instructed in SKILL.md to read company-context.md before responding. Additionally, the scripts comp_benchmarker.py and hiring_plan_modeler.py process external JSON files (e.g., roster.json, plan.json) provided via command-line arguments.
Boundary markers: Absent. There are no instructions or delimiters implemented to ensure the agent ignores instructions potentially embedded within employee names, business cases, or company context files.
Capability inventory: The skill possesses the capability to read local files and process sensitive human resources data. The scripts perform complex calculations and generate reports that are presented to the user.
Sanitization: Absent. The Python scripts parse JSON data directly into dataclasses without sanitizing or validating text fields for injection patterns or malicious instructions.

chro-advisor