The Agent Skills Directory

[SAFE]: The suite includes 25 Python analytical tools for financial modeling, organizational health, and strategic planning. All tools are implemented using only the Python standard library, which mitigates risks associated with unverified external dependencies or malicious packages.
[DATA_EXFILTRATION]: The skill implements a dedicated 'Anonymization Protocol' (context-engine/references/anonymization-protocol.md) that enforces the removal of sensitive identifiers, absolute financial figures, and specific names before any data is used in external searches or API calls. This architectural choice effectively prevents unauthorized data leakage during tool execution.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided company metrics and risk profiles through the onboarding and board meeting commands. While this constitutes a data ingestion surface for the processing of untrusted content, the risk is managed through structured data processing and the internal quality loop defined in the agent protocol.
[EXTERNAL_DOWNLOADS]: The documentation references installation via well-known package managers and the author's public repository. These references are for initial setup and do not involve runtime downloading of untrusted executable content.
[SAFE]: The 'Inter-Agent Protocol' defines a clear hierarchy and interaction model for multi-agent deliberation. It includes hard rules for loop prevention, such as a maximum invocation depth and circular call blocking, ensuring system stability during complex deliberations.
[SAFE]: The suite uses a local hidden directory for persistent context storage. Access to this directory is legitimate for the skill's primary purpose of maintaining company context across sessions, and no attempts to access unauthorized sensitive system files were observed.

c-level-advisor