The Agent Skills Directory

[Data Exposure] (HIGH): The skill is configured to read and synthesize extremely sensitive information, specifically data/cfo/cap_table.json (equity/ownership), data/cfo/latest_forecast.json (financials), and data/engineering/tech_debt.json (security/stability risks). Accessing these files centralizes the organization's most private data into the agent's context.
[Indirect Prompt Injection] (HIGH): The skill implements a massive ingestion surface by loading 12+ external JSON files across four departments. It lacks boundary markers or sanitization logic. If an attacker can influence any of these source files (e.g., via a compromised CRM or product roadmap), they can inject instructions to bias strategic outputs or manipulate the 'Weekly Leadership Brief' intended for senior executives.
[Dependency Risk] (MEDIUM): The skill explicitly relies on reading data directories managed by other skills (cmo, cfo, cpo, cto). This creates a complex trust chain where a vulnerability in a department-level skill or its data source automatically compromises the 'Leadership Sync' output.
[Command Execution] (SAFE): The skill does not contain any shell commands, subprocess calls, or runtime script execution patterns.

leadership-sync